A Dissertation Submitted to the Faculty of The

A Framework for Application Specific Knowledge Engines

Item Type text; Electronic Dissertation

Authors Lai, Guanpi

Publisher The University of Arizona.

Rights Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction or presentation (such as public display or performance) of protected items is prohibited except with permission of the author.

Download date 25/09/2021 03:58:57

Link to Item http://hdl.handle.net/10150/204290

A FRAMEWORK FOR APPLICATION SPECIFIC KNOWLEDGE ENGINES

Guanpi Lai

______

A Dissertation Submitted to the Faculty of the

DEPARTMENT OF SYSTEMS AND INDUSTRIAL ENGINEERING

In Partial Fulfillment of the Requirements For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

2010

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Dissertation Committee, we certify that we have read the dissertation

prepared by Guanpi Lai

entitled A Framework for Application Specific Knowledge Engines

and recommend that it be accepted as fulfilling the dissertation requirement for the

Degree of Doctor of Philosophy

______Date: 4/28/2010 Fei-Yue Wang

______Date: 4/28/2010 Ferenc Szidarovszky

______Date: 4/28/2010 Jian Liu

Final approval and acceptance of this dissertation is contingent upon the candidate’s submission of the final copies of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

______Date: 4/28/2010 Dissertation Director: Fei-Yue Wang 3

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at the University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: Guanpi Lai

ACKNOWLEDGEMENTS

I wish to thank my committee members who were more than generous with their expertise and precious time. A special thanks to Prof Fei-Yue Wang, my dissertation advisor and committee chair for his countless hours of reflecting, reading, encouraging, and most of all patience throughout the entire process. Thank you Prof. Ferenc

Szidarovszky, Dr. Daniel Zeng, and Dr. Jian Liu for agreeing to serve on my committee.

I especially thank Yanqing Gao, Yilu Zhou, Jialun Qin and many others for their encouragement and emotional support during my tough time.

DEDICATION

This dissertation is dedicated to my family: my wife Xuetao Xu, my child Lucas Luming

Lai, my parents Yangfu Lai and Chunrong Yang, and my parents-in-law Furong Xu and

Guiju Liang. I give my deepest expression of love and appreciation for the encouragement that you gave during this long journey.

TABLE OF CONTENTS

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

ABSTRACT ...... 11

CHAPTER 1 INTRODUCTION ...... 13 1.1 Motivation and Research Description ...... 14 1.2 Organization of the Dissertation ...... 17

CHAPTER 2 UNDERSTAND DATA ON THE WEB ...... 20 2.1 Two worlds of Data – Unstructured and Structured ...... 20 2.1.1 Manage unstructured data ...... 21 2.1.2 Structured data on the Web ...... 28 2.2 Life on the Web ...... 33 2.2.1 Online Communities ...... 33 2.2.2 Peer-to-Peer World ...... 41 2.3 Conclusions ...... 46

CHAPTER 3 A FRAMEWORK FOR APPLICATION SPECIFIC KNOWLEDGE ENGINES ...... 47 3.1 Knowledge Portals and Applications ...... 49 3.2 An Overview of the Framework for Application Specific Knowledge Engines… ...... 54 3.3 Construction of Data Repositories ...... 55 3.3.1 Data Collection ...... 55 3.3.2 Data Preparation ...... 71 3.3.3 Data Silo ...... 72 3.4 Searching by KCF with Result Presentation ...... 79 3.4.1 KCF Processing ...... 79 3.4.2 Semantic Search ...... 83 3.4.3 Result Presentation ...... 85 3.5 Conclusions ...... 89

TABLE OF CONTENTS - Continued

CHAPTER 4 SEARCHING TERRORIST GROUPS ON THE INTERNET…… ...... 90 4.1 Literature Review ...... 92 4.1.1 Digital Archiving for Terrorists’ Resources...... 94 4.1.2 Terrorism Research Portals ...... 95 4.1.3 Multilingual Issues ...... 97 4.2 Research Questions...... 98 4.3 Implementation of Dark Web Portal ...... 99 4.3.1 Dark Web Data Collection Building ...... 99 4.3.2 Post-retrieval Analysis and Multilingual Support ...... 115 4.3.3 Searching and Browsing in the Dark Web Portal ...... 119 4.3.4 Multilingual Support ...... 124 4.3.5 Semantic Search in the Dark Web ...... 126 4.4 Conclusions and Future Directions...... 130

CHAPTER 5 MONITOR FILE SHARING IN P2P WORLD ...... 132 5.1 Literature Review ...... 133 5.1.1 P2P History ...... 133 5.1.2 P2P Networks ...... 136 5.1.3 Related P2P Research...... 141 5.2 Implementation of Building ASKE Data Collection ...... 143 5.2.1 Resource Identifier ...... 143 5.2.2 Spider agents ...... 145 5.2.3 Content Filter...... 149 5.3 Services and Case Study ...... 152 5.3.1 Services for Copyright Owners ...... 152 5.3.2 Case Study – Watchmen ...... 157 5.4 Conclusions ...... 161

CHAPTER 6 CONCLUSIONS AND FUTURE DIRECTIONS ...... 162 6.1 Conclusions ...... 162 6.2 Future Directions ...... 163

TABLE OF CONTENTS - Continued

APPENDIX A US DOMESTIC EXTREMIST GROUPS AND URLS .... 165

APPENDIX B INTERNATIONAL TERRORIST GROUPS AND URLS FOR ARABIC GROUPS ...... 178

APPENDIX C US DOMESTIC EXTREMIST FORUMS ...... 202

APPENDIX D PART OF SOURCE CODES OF EDONKEY SPIDER AGENTS ...... 212

APPENDIX E PART OF SOURCE CODES OF BITTORRENT SPIDER AGENTS ...... 223

APPENDIX F SOURCES FOR WATCHMEN IN EDONKEY NETWORKS AND BITTORRENT NETWORKS ...... 235

REFERENCES ...... 265

LIST OF TABLES

Table 2.1 Profile of the types of analytic users in a typical organization ...... 22 Table 3.1 Popular Forum Software Packages ...... 66 Table 3.2 Different Types of Download Servers ...... 67 Table 4.1 Current Approaches to Archive Terrorists’ Web Resources ...... 93 Table 4.2 Number of pages collected for the terrorist groups within each category ...... 107 Table 4.3 Documents spidered in the second batch for US Domestic Extremist Groups ...... 108 Table 4.4 Documents spidered in the second batch for Arabic-Speaking Terrorist Groups ...... 109 Table 4.5 Documents spidered in the second batch for Spanish-Speaking Terrorist Groups ...... 110 Table 4.6 Comparison between the 1st and 2nd batch collection ...... 110 Table 4.7 Summary of Document Types in the Forum Collection ...... 112

LIST OF FIGURES

Figure 2.1 The screenshot showing the main forum discussing page ...... 34 Figure 2.2 The screenshot showing the threads in a specific forum discussion ...... 35 Figure 2.3 How web users participate in online community? ...... 36 Figure 2.4 Part of torrent file list for “Avatar” on the thepiratebay.org ...... 42 Figure 2.5 The screenshot downloading “Avatar.3D” using BitTorrent ...... 43 Figure 3.1 The framework and components of an Application Specific Knowledge Engine ...... 56 Figure 3.2 The structure of Resource Identifer ...... 58 Figure 3.3 The architecture of social sensor networks ...... 61 Figure 3.4 Ontology-development process ...... 75 Figure 3.5 The architecture of Metadata Extractor ...... 76 Figure 4.1 Number of U.S. domestic groups identified from each source within each category ...... 104 Figure 4.2 Number of international groups identified from each source within each geographical location ...... 104 Figure 4.3 The social sensor network for U.S. Domestic Extremist Groups ...... 106 Figure 4.4 Summary and categorization of identified U.S. domestic extremist forums . 111 Figure 4.5 An example of U.S. domestic extremist forums ...... 114 Figure 4.6 The distributions of number of participants and number of postings on extremist Forums ...... 115 Figure 4.7 The screenshots of Dark Web Portal for U.S. domestic groups ...... 127 Figure 4.8 The screenshots of Dark Web Portal for Arabic-speaking terrorist groups .. 128 Figure 4.9 Jihad multilingual portal user sessions ...... 129 Figure 4.10 The screenshot of semantic search for the keywords "Roubaix Gang" ...... 130 Figure 5.1 The Architecture of P2P Spider Agents ...... 147 Figure 5.2 Customize user profile ...... 154 Figure 5.3 Infringement overview page ...... 155 Figure 5.4 Infringement search interface ...... 155 Figure 5.5 Top 5 countries of peers and trackers for “Quantum of Solace” between 11/03/2008 to 11/21/2008 ...... 156 Figure 5.6 Protocols breakdown ...... 158 Figure 5.7 Daily infringements trend – all infringements...... 160 Figure 5.8 Top 10 ISPs ...... 160 Figure 5.9 Top 10 filenames ...... 161

ABSTRACT

The amount of information on the Internet has been proliferated rapidly in recent years as new technologies and applications become popular. The broad heterogeneous contents bring us a substantial challenge in the field of knowledge discovery and information retrieval. The objective of this dissertation is to design and implement a systematic framework to help users access huge and various information on the Web by combining different techniques and algorithms in different domains. In this dissertation, we propose an effective Application Specific Knowledge Engine framework to build structured and semantic data repositories, and support keyword search and semantic search. The framework is consistent with the architecture of most search engines. It enhances the general search engines in three ways: various data retrieval ability; semantic data support; and post-retrieval analysis. Various techniques and algorithms that could facilitate knowledge discovery are used in the framework.

In the first part, we review different types of data on the Internet and approaches to retrieve various data: structured and unstructured data, online community data, and Peer- to-Peer data. After that we present an overview of the system architecture of the ASKE framework, and especially discuss the core components of the framework in details.

The following chapters aim to investigate how the ASKE framework can be applied in two different domains (counter-terrorism and anti-piracy). We present the research in developing a counter-terrorism knowledge portal that incorporates various data collection

and post-retrieval analysis. The process of building the portal following ASKE framework is described. The details of the data collections of Web sites and online forums are also reported. In the anti-piracy domain, we mainly discuss building Peer-to-

Peer data collection and serving users with customized profiles. A case study of monitoring the movie Watchmen piracy on typical Peer-to-Peer Networks is discussed also.

This dissertation has two main contributions. Firstly, it demonstrates how information retrieval, Web mining and other artificial intelligence techniques can be used in heterogeneous environment. Secondly, it provides a feasible framework which can facilitate users to discover knowledge in their specific searching and browsing activities.

CHAPTER 1

INTRODUCTION

In this dissertation, we demonstrate that effective and efficient knowledge discovery has become a more and more challenging problem, as the massive amount of Web pages and data being added to the Web every day. People have published a large number of academic papers on knowledge discovery with the continually increased interest, which can be attributed to information overload and data explosion, the advances in information technologies, and the need for organizations and individuals who are expected to keep their competence, although most of them target to solve a specific part of the problem.

We propose a highly integrated framework, called Application Specific Knowledge

Engine [1] (ASKE ), for intelligent interactive information retrieval and knowledge discovery from the Internet, and exploit different kinds of existing technologies such as data mining, text mining, and web mining, also include the related research topics with

Web spidering and general search engines. We also illustrate how this framework is applied to different applications (Dark Web [2] and Peer-to-Peer World).

A 1998 report [3] on database research foretold that the Internet will hold the majority of published human knowledge. With the advanced Web technologies (e.g. Web 2.0), never has it been easier than with the Internet to publish all kinds of digital materials and make them instantly available to anyone. However, as the amount of data grows exponentially, availability extends to the issue of universal accessibility. It becomes more and more

necessary for us to seek alternative methods to effectively and efficiently discover and use the invaluable assets hidden in huge amount of data.

As text search on search engines become routine as millions of users use them daily to pinpoint resources on the Internet, knowledge discovery capability on the Internet is still limited and sometimes even absent when simple keyword searches can bring back hundreds of thousands of document results. The most popular keyword indexing techniques in most of the current information retrieval systems available on the web emphasize very little on context or textual information, which is actually very important during the knowledge discovery process. It’s imperative to break the limitation of simple keyword searches and take account into implicit semantic information. In addition, search engines lack the ability to unveil the details of people’s activities on the Internet (e.g. downloading movies/music using different applications and protocols, chatting in open chat rooms, posting posts on forums, etc.)

1.1 Motivation and Research Description

We motivate the dissertation by discussing a number of open problems in the previous section.

1. Information Overload in the World Wide Web space

"Information overload" is a term referring to an excess amount of information being provided, making processing and absorbing tasks very difficult for the individual because sometimes we cannot see the validity behind the information [4], created by Alvin Toffler.

In the World Wide Web (WWW) space, more and more people are not only the consumers of data but also producers. As more and more new technologies are developed, it becomes much easier to retrieve, produce and deliver information than in earlier period, and thus all relevant or irrelevant data is published instead of only the most important one.

This results the explosion in more and more unclear and inaccurate information on the

Web. Finding real useful information is often a hit-and-miss process.

Using searching engines to find data in WWW space is a well-known approach, however, almost everyone would agree that extracting information from Web via the search engine technology is still a skill that not many people are able to master. Widely used search engines tend to return thousands of answers for even very specific queries, which do not have any kind of underlying semantic structure that could facilitate knowledge discovery.

2. Diversity of Information Sources

It becomes a big problem that valuable information reside in diverse sources, as today we lack the mature methods to locate, access and integrate various information. As people and organizations rely more and more on the Web for their daily lives and daily operations, the data on the Web evolves into huge diversity in both locations and formats.

There are two issues are involved. The first one is how to collect diverse information from sparse locations efficiently, and second is how to store the data and integrate them.

It usually creates large, rigid, and practically unmanageable systems by tightly integrating diverse sources into a single system to centralize the information sources.

Using federated systems [5] may be an alternative. However, a central data management dictionary is required to translate search queries into terms/words that the diverse information sources can understand in most federated systems. The federated systems have to update the dictionary frequently, thus make the entire data sources unusable and unavailable.

3. Data Mining and Text Mining

A lot of data is stored today so that people can discover valuable knowledge and use them during the decision making process. Data and text mining tools are required to extract useful information from large amounts of diverse information sources.

Generally, data mining, also called Knowledge Discovery in Database (KDD), is the process of analyzing data from different perspectives and summarizing it into valid, novel, potentially useful, and ultimately understandable information - information that can be used in decision making from different types of data storages, such as databases, data silos, etc. It allows users to analyze data from many different dimensions or angles, categorize it, and summarize the relationships identified. Technically, data mining is the process of finding correlations or patterns among dozens of fields in the data that are organized in records structured by categorical, ordinal and continuous variables, such as large relational databases.

However, most of data that are stored today are actually unstructured. Up to 90% of all organizational data is stored in some sort of unstructured text [6] according to a recent study by Merrill Lynch and Gartner,. Text mining is the process of deriving novel

information, such as relations, hypotheses or trends that are hidden in a collection of unstructured text files. It numerates and characterizes the unstructured documents into structured table representation which can be explored by data mining tools. How to utilize and integrate data mining and text mining technologies to discover knowledge from a high volume of unstructured data from the Internet becomes a challenging issue.

4. Data Repository

The Internet is a collection of objects with complex structures. These objects can be physical, like web pages, documents, images, videos, sound, maps, games, applications, data files etc. They can be also virtual, like users, hosts, network etc., as some roles on the Web. These objects are distributed and stored, or represented, on a large set of heterogeneous repositories. It’s very costly to query such data due to the huge semantic ambiguities and the heterogeneity of the data sources.

One way to reduce the cost is by storing the processed data in a relational table at a more general conceptual level[7]. Users may scan the general description of the information on the Internet by applying high level queries directly on the processed data. Another way is to develop a semantic-web-based data repository by using the Semantic Web technology, which provides consistent formats to integrate and combine of data which is from diverse sources and language for representing the relationships between the data and real world objects.

1.2 Organization of the Dissertation

Traditional search techniques cannot satisfy people’s needs for knowledge discovery from heterogeneous data on the Internet. To address this issue, we propose the following research questions:

1. How can we develop a generic framework to facilitating searching and browsing

on the Internet by integrating various data collection techniques and post-retrieval

analysis from heterogeneous data?

2. How can we build structured data repositories and semantic data repositories

systematically and efficiently?

3. How can we apply the generic framework to different applications and domains?

In Chapter 2, we present the research on different types of data on the Internet, and approaches to retrieve heterogeneous data: structured and unstructured data, online community data, and P2P data.

Chapter 3 demonstrates the concepts of ASKE and proposes a feasible approach in many other different applications. We present an overview of the system architecture of the

ASKE framework, and explain the core components of the framework.

In Chapter 4, we present the research in developing a counter-terrorism knowledge portal that incorporates various data collection techniques and post-retrieval analysis. The process of building the portal following ASKE framework is described. We also report the data collections of Web sites and online forums.

In Chapter 5, we mainly discuss how to monitor P2P networks using ASKE framework.

Different P2P protocols and current research topics in P2P networks are reviewed. The detail of implementation of building data collection is presented. We also report the Anti-

Piracy system which uses the P2P data collection to serve users with customized profiles.

A case study for monitoring the movie Watchmen piracy on eDonkey network and

BitTorrent Network is discussed.

Chapter 6 summarizes conclusions of this dissertation and suggests future directions.

CHAPTER 2

UNDERSTAND DATA ON THE WEB

Today’s Web has millions terabytes of information available to humans, but not accessible to computers. Most of the data on the web reside in HTML pages, which are formatted in esoteric ways that are difficult for machines to process and to locate for humans. In addition, the Web is not only comprised of millions of web pages but also people’s actives as Web 2.0 techniques and social network become more and more popular. To discover knowledge from the huge data, it would be necessary to understand the data in the ways of structures, formats, accessibilities, and activities.

In this chapter, we explore the characteristics of data on the Web, and review the ways to approach the problems that are the results of these features.

2.1 Two worlds of Data – Unstructured and Structured

Since the Web emerged, unstructured content has dominated the Web. The techniques of

Information Retrieval are applied broadly in searching the Web. Particularly, it's a truism that more than 85 percent of business-relevant information exists as unstructured content, such as e-mails, reports, letters, surveys, white papers, memos, news, user groups, chats, marketing material, presentations and Web pages. This figure (85 percent) is very widely cited by analysts and vendors [8].

People use unstructured data on the Web every day, creating, storing, delivering and retrieving web pages, reports, e-mails, spreadsheets and other types of documents.

Unstructured data consists of any data stored in an unstructured format at an atomic level.

There is no conceptual definition and no data type definition in the unstructured content.

In textual documents, a word is simply a word. Most of unstructured data are textual objects, which are based on pre-defined platforms or media, such HTML, Microsoft

Word documents, e-mails, presentation slides or spreadsheets.

In the mean time, people also use structured data a lot. Structured data usually has an enforced form composed of different atomic data types. Structured data supports querying and reporting against predetermined data types and understood relationships, as structured data is organized in a highly regular way, such as in tables and relations, where the regularities apply to all the data in a particular dataset [9].

2.1.1 Manage unstructured data

To manage unstructured data efficiently is considered as one of the major challenged problems in the information technology research areas. White-collar workers spend anywhere from 30 to 40 percent of their time managing documents, up from 20 percent of their time according to investigations from Gartner. The main reason is that the techniques which can successful transform structured data into business intelligence simply don't work with unstructured data, as they relies much on the structure of data.

Compared with structured data, three distinct challenges are raised for unstructured data:

1 Although a lot of unstructured data does formatted in some forms, such as

HTML, XML or Microsoft Word templates, the data is still not accessible from

a semantic level.

2 We cannot gain insight from unstructured data as the context of the information

is missing.

3 And lastly, the semantic meaning of information is interpreted largely

subjectively instead of objectively.

To understand the extension of this problem, Giga Research conducted a study to profile the types of analytic users in a typical organization. In Table 2.1, the rightmost column shows that all classes of users spend considerable amounts of time using decision support tools and applications. From this table, we can notice that the vast majority of decision makers in an organization are information consumers, and typical business intelligence solutions are generated based on structured data alone thus satisfy decision makers requirements.

Type of Users % of All % That are % That are Decision % of Work End Users Producers Consumers of Makers Time Data/Analysis Information Allocated to Business Intelligence Solutions IT 2 98 2 No 15 Power User 5 84 16 No 42 Business User 25 18 82 Yes 12 Casual User 30 8 92 Yes 4 Extended 38 3 97 Yes 2 Enterprise User Table 2.1 Profile of the types of analytic users in a typical organization

The common approach to manage unstructured data is to convert them into structured data so that all the efficient tools and techniques which are useful for structured data can be directly applied. Search engines play very important roles during this process.

1. Identify and Locate Data - Search

The first step to manage unstructured documents is to make them searchable. Prior to the emergence of the Web, full-text and other text-search techniques were mainly applied within library, document-management and database management systems. However, with the exponential growth of the Internet, searching on the Web quickly became the main way to retrieve information. According to market-research firm Outsell Inc., office workers now spend an average of 9.5 hours each week searching, gathering and analyzing information; and nearly 60 percent of that time, or 5.5 hours a week, is spent on the Internet, at an average cost of $13,182 per worker per year [10].

All this searching is not efficient as we expected. Current Web search engines are implemented similarly to traditional information-retrieval systems: indexes of keywords within documents are generated and stored in database systems, and when a user query is received, the query is split into several keywords which are searched in the database system, finally a ranked list of documents is returned. The quality for search results relies much on the input user queries which are out of control of search engines. In fact, the average length of search terms used on the public Web is only 1.5 to 2.5 words, as several studies show. In addition, the average search contains efficient Boolean operators (such as and, or and not) is fewer than 10 percent of the time. With such short and ambiguous

queries and lack of advanced search techniques, the results are not good enough. A performance assessment of the top five Web search engines, conducted by the U.S.

National Institute of Standards and Technology, showed that when 2.5 search words are used, only 23 to 30 percent of the first 20 documents returned are actually relevant to the query.

The search-engine vendors have continued to improve their technology with the acknowledgement of the weakness of basic keyword search. For example, Google has added techniques such as stemming and spelling correction, time and category classification to its widely used search engine, while Powerset employs natural language processing [11].

2. Search in Context

Another problem of Web search engines is that they don’t take account of the context of data, as they treat each search request independently. Although users’ interests differ, if they input same queries, the results will be identical for every user. For example, for the query “smart phones", the first page of results for a recent query to Google is the

Wikipedia page about “smart phones”, introducing the history, operating systems etc. about “smart phones”. This is useful information if the user is interested in the idea of what smart phones are, but not for a buyer interested in shopping a smart phone. If the context for the query is known to be that the user is buying a smart phone, it is possible to refine the query accordingly via knowing the type of result in the first place.

Adding contextual information would help improve search efficiency a lot. Contextual information generally is meta data, and it helps narrow down the range of possible results.

It can be also extracted from carefully-crafted set of indexed resources. "Parametric selection" allows users to locate and access data by providing a way to utilize available meta data or context description [12] by filtering and sorting on known meta data fields such as geographic locations, time, product codes, topics etc. By carefully selecting only the relevant fields, the user can tremendously reduce the number of results that may be more relevant to the search.

To construct an "advanced search" form is another popular approach. Users can select options on meta data to specify their search in more details. For example, Atomz's

Content Mining Engine believes that meta data-assisted search overcomes many of the limitations of standard keyword search [13].

3. Classification and Taxonomy

Structured data is usually organized as rows and columns, such as in spreadsheets or relational databases. Similarly, we can manage unstructured data in systems with a hierarchical structure, which is called “taxonomy”. A taxonomy provides users a convenient, intuitive way to explore and access information, similar to the file directories in Microsoft Windows systems. The users can drill down through the categories and subcategories of the taxonomy until they find the relevant data or documents, rather than formulating a query and then go through the ranked results. In addition, the taxonomy can

be used as meta data in advanced search forms to limit queries by specifying categories and sub-categories.

The process of organizing or converting unstructured data into a taxonomy is referred to as classification, which is a widely used technique in various fields, including data mining [14]. It classifies a large set of objects into predefined classes, described by a set of attributes, using different types of supervised learning methods. Classification learning algorithms have been extensively implemented in machine learning with quite a long history. Pure symbolic machine learning algorithms are the most common ones, although some of them are specialized in particular types of data and domains, which are very useful in specific domain knowledge retrieval. There are many different types for classification algorithms: a) rule based algorithms, CN2 [17] and CL 2 [18]; b) decision tree algorithms, ID3 [15] and C4.5 [16] etc.; c) pure statistical algorithms, CART [19], genetic algorithms [20], adaptive spline methods [21] and graphical models [22]; d)

Nonlinear algorithms based on neural networks, back-propagation networks [23], and nonlinear regression; e) example-based algorithms, PEBLS [24], algorithms based on inductive logic programming(e.g. [25, 26]) and hybrid systems (e.g. [27]); f) others,

Support vector machines [28, 29], conformal predictors [30] and Bayesian classifiers [31], etc.

Although many people agree taxonomy/classification could be the standard way of processing unstructured data, there are couple stumbling challenges. One major issue is the unacceptably low accuracy results of automated classification systems. These systems

usually only use a single technique, such as a rule-based or a example-based approach as we mentioned above, which works fine in some data or domains, but not in others. For a particular data set, we have to test different algorithms, select the best fit, and tune it up.

The difficulty of maintaining taxonomies would be another challenging issue. This task requires individuals who both understand domain knowledge and happen to have rich IT experience. This kind of human resources is hardly to find and would cost a lot. In addition, a high quality taxonomy with broad coverage may be at least eight to 10 levels deep and contain hundreds, even thousands, of categories. Since these taxonomies are based on dynamic data, such as techniques, products, company organizational structures, or financial data etc., they need to be modified very frequently. Updating and maintaining such taxonomies are not an easy job, both time-consuming and expensive.

4. Content Intelligence: Toward a Solution

Search engines are very helpful to locate unstructured web pages via keywords or simple logic combinations of keywords. The common form that search results are presented is a list, and each search result (URL/link) is independent. A new enhanced search and intelligent classification system are expected. This system will provide intelligent services that generate incremental value for users.

The solution “content intelligence” would be fully developed applications that are not limited to search and document classification. These applications could help users navigate and access their unstructured data, extract meta data from documents, classify the documents, build up a taxonomy, and provide a sophisticated user interface for

browsing the documents’ hierarchy. In addition, through system interfaces, such as APIs or XML-based Web services, applications can be isolated from the underlying shaggy information hierarchy, which may change very often.

Content intelligence is supported by the following technologies: search, classification and discovery. Discovery here describes the process of automatically searching large volumes of data for patterns that can be considered as the knowledge about the data. Content intelligence will be used to uncover new issues and trends and to answer specific business questions, akin to business intelligence, thus unstructured data on the Web will become a source of a source of valuable, accessible, time-critical knowledge repository and business intelligence.

2.1.2 Structured data on the Web

Although the Web is dominated by unstructured data for a long time, we can see a huge increase both in the amount of structured data on the Web and in the diversity of the structures in which these data are stored. Most recent examples of structure data are various annotation schemes (e.g., Digg [32], Flickr [33], Twitter [34]) that allow users to add labels/tags to content, like pages or images. A majority of structure data can be found in the deep Web (also called the invisible Web or the hidden Web), which is referring to dynamic contents or pages on the Web that are stored in databases and served by a submitted query or accessed only through a HTML form. In the following section, we describe these two kinds of structured data that exists on the Web today, and discuss the level of structure for each kind of data.

1. The Deep Web

The deep Web contains the Web’s information that is buried far down on dynamically generated sites, and standard search engines do not find. Traditional search engines cannot retrieve content in the deep Web until those pages are created dynamically as the result of a specific query request is sent to the website. This content is considered invisible because web crawlers rely on hyperlinks to discover new content, and crawlers do not have the ability to fill out arbitrary HTML forms or generate dynamic queries.

Even web crawlers can get dynamic content, the result data doesn’t contain the structure nature which is built in the backend databases.

No one knows the size of the deep Web for sure. In 2000, Michael Bergman estimated that the deep Web contained approximately 7,500 terabytes of data and 550 billion individual documents [35]. Danny Sullivan, a search engine expert and formerly of

Search Engine Watch, wrote in 2000 that the invisible Web was about 500 times Google's index of one billion pages [36]. New estimates of Google's index sets it at over 8 billion in 2005 [37], and search engines are said to only crawl 16-20% of the Internet [38]. The content on the deep Web is believed to be possibly much more than the current indexed

WWW, and typically is of very high-quality [35].

To understand how much deep web content exists, there is no way to just count the number of forms on the Web. For instance, there are numerous applications and forms on different types of web sites, but actually utilize the same underlying data source, e.g. databases. As most of major search engines provide codes to put their search forms on

any web sites, a large number of forms on the web are simply the search boxes from search engines. Online communities and Web 2.0 applications are another two large sources of forms. People contribute user specific data, like posts, comments, blogs to these web sites. In addition, forms are also used for applications such as logins, subscriptions, comments, etc. Looking into these different types of forms, we can find that none of which present useful structured data to end users directly.

The content on the deep web cover a wide range of different categories. The deep web contents may cover: a) geographically specific information, such as locators for chain stores, businesses, and local services (e.g., doctors, lawyers, architects, schools, tax offices); b) reports with statistics and analysis generated by governmental and non- governmental organizations; c) product and price search, such as Google shopping, Bing cashback program, PriceGrabber, etc.; d) a variety of other data, such as art collections, public records, photo galleries, bus schedules, etc. In fact, deep web sites can be found under most categories of the ODP directory [39]. These sources can be accessed by simple keyword queries through a single search textbox, or detailed forms with many select-menus and other refinement options. In addition, as online communities/forums become universal for every subjects/topics, the content on these forums covers more widely.

Three nice properties of deep web content are of high quality, very specific in nature, and well managed. Consider CiteSeer [40] as an example. Documents stored in CiteSeer are authored by professional writers and published in professional journals or conferences.

They focus on very specific topics or domains. Weather.com provides a national and local weather timely and specific forecast for cities, as well as weather radar, report and hurricane coverage. Both collections share the three properties.

The deep web is extremely popular on the current Internet, and it yields much more valuable content than the surface web. Deep web content can easily provide many different types of valuable services, such as Online TV guides, price comparison websites, locating cheap or used textbooks, driving direction guiding sites, tracking the prices of your stocks and news about companies you are interested in.

2. Annotation Schemes

Annotation schemes are now popular in Web 2.0 applications or social networking websites, such as Digg, Twitter etc. They enable users to add tags describing underlying content (e.g., photos, comments, news) to enable better search over the content. The

Flickr Service by Yahoo! is a prime example of an annotation service for photo collections. Ahn and Dabbish [41] took this idea to the next level by showing how mass collaboration can be used to create high quality annotations of photos.

In a sense, annotation schemes require the users to provide very minimal structure, specifically, only the annotations, which provide the values for “About” to every piece of content. Obviously, annotations can be incorrect and inconsistent, i.e., use different words to describe same piece of information.

3. Access and Spider Structured Data

The problems faced in the context of annotation schemes to access and collect data are somewhat simple. Typically, the annotations can be used to improve recall and ranking for resources that have very little meta-data associated with them (e.g., photos, videos). It would be easy to solve the issue to discover deep web contents if structured data owners can just make the data public on the Web. Currently only a few high-profile sites like

Amazon or YouTube try to provide public Web services or custom application programming interfaces that open their databases and many more sites do not.

Currently there are no automatic ways about how to fill out any arbitrary forms. Actually a lot of human efforts have to be involved to determine what information a particular web form need. By analyzing forms’ elements and preparing set of possible values for each element, we can interact with the web server which answers the requests of the form, and send it the information that specifies the query plus other data the web server needs. We can imagine that there are many of different types of forms. It’s almost impossible for search engines to search them all, as a lot of human intervention is required during the process.

While filling out that web form is difficult to handle, it isn't the only challenge to accessing the deep web and it isn't even the toughest problem. It’s much harder to finding the best, or most relevant, content. We will have to search multiple sources, collate the results, remove duplicates and sort the remaining results by some criteria that is meaningful to the user in some specific domain.

There are enormous amounts of valuable structured data on the Web today, and we will finally address the challenges of making such data available and combining it with unstructured data.

2.2 Life on the Web

Many of people now spend much time at social-networking sites like Facebook,

MySpace or LinkedIn. It is pretty easy to happily consume time in browsing, posting, commenting on online communities. More and more activities in real life now are virtualized on the Web: shopping, job hunting, dating, watching movies, listening to music, seeking legal advice, gaming, chatting with friends, discussing big events, etc.

People even start to look for ways to abandon the virtual life and get actual life back, like

Web 2.0 Suicide Machine [43]. In this section, we will discuss online communities and how people use latest peer-to-peer technologies to share resources, such as movies, music, books, etc.

2.2.1 Online Communities

Based on Horrigan’s research [44], at least eighty-four percent of Internet users have contacted or participated in an online community, and the growth in membership and usage is expected to continue [45]. The common perception of online communities is that they allow groups of people to share ideas and information. An online community can be defined as an aggregation of individuals who interact around a shared interest or topic, where the interaction is at least partially supported and/or mediated by technology and

guided by some protocols or norms, such as online forums. Figure 2.1 and 2.2 show the screenshots of the main forum discussion page and the threads in a specific forum discussion.

Figure 2.1 The screenshot showing the main forum discussing page

Although everyone can contribute an online community equally in theory, the vast majority of online conversation or posts are conducted by a small group of web users, which is less than ten percent of them. The rest of the community quietly sits back and

Figure 2.2 The screenshot showing the threads in a specific forum discussion

listen interactions as a mostly-passive audience that only occasionally injects a few comments. Jakob Nielsen describes this phenomenon as the “90-9-1 rule” in

“Participation Inequality: Encouraging More Users to Contribute” [46]. It states:

• 90% of users are lurkers (i.e., read or observe, but don't contribute).

• 9% of users contribute from time to time, but other priorities dominate their time.

• 1% of users participate a lot and account for most contributions: it can seem as if

they don't have lives because they often post just minutes after whatever event

they're commenting on occurs.

Rubicon’s survey confirms the idea behind the “90-9-1” rule, although it does not support its specific details, as Figure 2.3 [47]. A majority of web users sometimes contribute a little bit even if it’s just an occasional comment. The truly silent lurkers are only 9% of the web population. But the vast majority of content still comes from a small group of the population.

Figure 2.3 How web users participate in online community?

There are several attributes that can help us to characterize online communities, including

Purpose, Platform, and People [48].

• Attribute 1 – Purpose: The notion of purpose is the key to a virtual community as

the community itself is defined by shared purpose among community members.

People in the community understand and buy into its mission; they have a shared

vision for what it is, what it could become, and where it’s heading. They work

together to make progress toward a common goal.

• Attribute 2 – Platform: the communication among members of an online

community is synchronous or asynchronous. For example, chat room technology

supports real-time communication (i.e. synchrononous interaction) whereas

email-based forums allow members to view and respond to messages at their

convenience rather than in real time (i.e. asynchronous interaction).

• Attribute 3 – People: The people who interact with each other in the community

and who have individual, social and organization needs. Some of these people

may take different roles in the community, such as leaders, protagonists,

comedians, moderators, etc. on the earth, online communities provide channels or

platforms for people to communicate and interact online, thus the behaviors and

pattern of interaction become the most important attribute of a given online

community.

• Attribute 4 – Policy: The language and protocols that guide people's social

interactions and contribute to the development of folklore and rituals that bring a

sense of history and accepted social norms. More formal policies may also be

needed, such as registration policies, and codes of behavior for moderators.

Informal and formal policies provide community governance.

In a thriving online community, people would like to share histories or share experiences in the workplace, in their educational or vocational backgrounds. They may have worked at the same company or attended the same school, or have the same professional credentials, for example all those groups that are springing up within the LinkedIn environment. Members feel a sense of belonging, feel “at home,” welcomed – free to speak and act in ways that are authentic to who they really are. Members often have shared values. The community exerts a magnetic attraction (may grow virally), and pulls new like-minded people into its gravitational force field, who in turn attract others.

Online communities help companies to communicate directly with their customers and gain valuable insight into what consumers think. There are several ways for companies to use an online community.

1. Get customers involved in the business

Customers are likely to have extensive experience of the products, and how they are actually used before they pull the trigger to get the products. A company would like to check with the people who know its products best when a company plan to commercialize a new product or idea, find out how people are using the product or find out about how the company is different to the competitors. An online community can provide a very appropriate platform between producers and customers, acting as a customer voice and empowering internal teams with customer input and insight.

Feedbacks can be timely got from the online community which is able to represent the customer inside the business.

2. Innovate with customers

In addition, an online community provides two way communication, and can be a great way of both getting new ideas organically, and of working with customers on innovation and co-creation. New ideas or features of products can be generated during ongoing discussions between customers and producers by involving customers throughout the innovation process, rather than just testing ideas at specific stages. An online community makes it possible for internal experts and others with customers to talk with each directly, and bring to the innovative ideas which might never have thought of before by experts themselves.

3. Find out how customers interact

Customers are usually treated as isolated beings who can answer questions about their habits and behaviors in traditional market research. It has a big problem as it doesn’t consider the social context which is the most important aspect of any market decision.

Via an online community it is easier to observe the conversations people have. By observing what they do and how they talk to their peers, we can figure out how they discuss and evaluate products or competitor products, e.g. pros and cons, how they advise other people, how they explain their decisions and opinions, what they choose to discuss.

4. Learn the language customers use

How to describe a product precisely and efficiently is a big challenge for many brands and products. An online research community can help you analyze and draw insight from the language people use.

5. Find answers to questions that the company didn’t even know to ask

Traditional market research usually designs a set of pre-defined questions to ask customers, but how about those questions that are related to customers’ opinions and are not ask in the traditional survey. The discussions in online communities will let the company understand what customers talk to each other about, what they really think about products and how they really talk about. It’s possible for the company to find out what matters to them most and what they think about it. And most importantly they will ask questions. Community members debate with each other, and thus generate ideas in a well managed online community.

There are a lot of obvious research topics that focus on online communities, including: the life cycles of different types of online communities; how online communities organize themselves and develop norms, rules and policies; how to discover people’s opinions on some specific topics or events automatically and efficiently; how to better support sociability and usability in online community development; how to evaluate and measure the success of different types of communities from participants’ and commercial sponsors’ perspectives.

These topics mainly address online communities themselves. As online communities become more and more popular, they actually are bridging online and offline social

networks, and people’s activities (such as opinions on events, comments on products, arguments with other people, etc.) in online communities become projections of their real life. For example, Facebook is used to maintain existing offline relationships or solidify offline connections, as opposed to meeting new people. Given that online communities enable individuals to connect with one another, it’s not surprising that they have become deeply embedded in user’s life. Boyd [49] argues that MySpace and Facebook enable U.S. youth to socialize with their friends even when they are unable to gather in unmediated situations; and these Social Network Sites are “networked publics” that support sociability, just as unmediated public spaces do.

2.2.2 Peer-to-Peer World

A peer-to-peer , commonly abbreviated to P2P, is any distributed network architecture composed of participants that make a portion of their resources (such as processing power, disk storage or network bandwidth) directly available to other network participants, without the need for central coordination instances (such as servers or stable hosts) [50]. Peers are both suppliers and consumers of resources, in contrast to the traditional client-server model where only servers supply, and clients consume.

P2P applications can be used for many purposes such as Internet Telephony (Skype [51]), distributed computing (SETI@home [52]) and content distribution (BitTorrent [53]).

Here we will focus on P2P applications used for content distribution because of their greater impact on carriers’ traffic, and P2P was popularized by file sharing systems like

BitTorrent. The basis of P2P applications is a group of peers that collaborate in a

distributed system for the purpose of distributing content to one another. This does not fit the traditional client-server model as each peer is a client and server at the same time [54].

Figure 2.4 Part of torrent file list for “Avatar” on the thepiratebay.org

According to the report from Mary Madden and Lee Rainie [55], about 36 million

Americans—or 27% of internet users—say they download either music or video files. 58% of those who were current music downloaders said they had specifically pulled music files from peer-to-peer services; 31% of these music downloaders said they were actively downloading music files from these networks and 27% said they had used this source for music in the past. When we isolate current video downloaders as a separate group, we find that 32% of them admit to ever using peer-to-peer to download either music or video files. Looking at current music and video downloaders in the aggregate, 33% report ever

using file-sharing networks to get music or video files. There is no doubt that peer-to- peer file sharing systems and applications play as a significant role in people’s life on the

Web.

Figure 2.5 The screenshot downloading “Avatar.3D” using BitTorrent File sharing peer-to-peer systems such as BitTorrent [53], Gnutella [56], and eMule [57] have significantly gained importance in the last few years to the extent that they now dominate Internet tra ﬃc [58, 59], way ahead of HTTP traffic. It has thus become of primary importance to understand how, when, who use these applications to download what kind of resources, which are people’s activities on the P2P network. Figure 2.4 shows the part of torrent file list for “Avatar” found on thepiratebay.org, and Figure 2.5

shows downloading “Avatar 3D” using BitTorrent, connecting to 30 seeders and 8 leechers.

In a P2P file sharing system, a distributed group of peers exchange contents in a decentralized manner. Peers act both as servers and clients, providers and consumers. The challenge in such system is how to help peers to find interesting content. Current popular

P2P file sharing systems usually offer search capability with the most usable indexing functionality. There are three popular ways to provide search functions helping peers locate the contents they are interested in. Firstly, like search engines, eMule [57] maintains a series of centralized indexing services although the files themselves are distributed. Another implementation is to broadcast the search requests over the network connected peers, as does Gnutella [56]. Lastly, new P2P applications, such as Kazaa [60], provide a kind of hybrid solution that does not have a single central index but kind of

"super nodes" that contain the index of content available in the proximity of the super node.

To get the download completed as soon as possible, all the current P2P applications open and sustain a large number of connections for each file downloaded and try to use as much bandwidth as is available. This common feature results unbalance among the users of an ISP. There is no easy solution to this issue as P2P applications transfer the data in different ways.

There are several challenges to monitor the whole P2P network:

• Difficult to locate download links: P2P network is world-wide range, and it’s not

like web sites or web pages which can be identified and located via URLs;

• Network connection limit: even given a specific download link, for example an

eMule link “ed2k://|file|.Saving.Private.Ryan.1998.720p.BluRay.x264.DTS-

WiKi.mkv|13238649556|fbabcf591bf0e6403e1ab8a4fba7cac1|h=wdouxamsn46fx

65si4sxsjc5oxialvpt|/”, we cannot monitor all the peers that download or share

this link because of our network connection limit, thus result incompleteness;

• Content Attack (Poisoning and Decoy): in one hand, more and more copyright

holders, such as movie studios, have been investing a lot of resources to

investigate technological solutions to prevent distribution of copyrighted materials

in p2p file sharing networks, and a popular technique that is applied now is

“poisoning” a specific item (movie, song, or software title) by injecting a massive

number of decoys into the p2p network, in the other hand, anyone can create fake

files and inject them into the p2p network; so even after we identify a peer to

download a specific item, we cannot assert it’s downloading the real contents.

• IP Spoofing: it creates TCP/IP packets using somebody else's IP address. Routers

use the "destination IP" address in order to forward packets through the Internet,

but ignore the "source IP" address. In this case, malicious peers can tempt to form

file-sharing clusters with known user IPs, resulting that we may claim the known

user IP downloaded copyrighted contents wrongly.

2.3 Conclusions

In this chapter, we go through different types of data on the Web, explore the characteristics of these types, and review the ways to approach the problems that are resulted by these features. Most of the data on the Web are unstructured, and a lot of researchers and companies have already provided methodologies and techniques to make them structured and accessible. Although structured data on the web is not that easy to access as unstructured data, because of the high value they are worth digging and developing. Online communities become the major platforms for people to have social life virtually, and they become another virtual society providing worthless materials for people to research. P2P file sharing systems give users a fast and efficient way to switch and broadcast contents. By monitoring p2p network, we can easily understand how, when, who download what kind of resources, which are people’s activities on the P2P network.

CHAPTER 3

A FRAMEWORK FOR APPLICATION SPECIFIC KNOWLEDGE

ENGINES

As the enormous and yet exponentially increasing amount of web pages have been distributed over the Web, it becomes more difficult to locate the exact information that a user may be looking for due to the intrinsic nature of dynamic and unstructured web information organization. A popular approach to overcome this problem is to use search engines such as Google and Yahoo! which crawl web sites, download web pages and create corresponding word/phrase indices. Although these general-purpose search engines help people to unearth the relevant information from billions of documents easily to certain extent, such search engines still bring some inevitable problems.

The direct problem is that the general-purpose search engines may often return a vast amount of documents as a result of retrieval which includes many irrelevant web pages and little indication as to which document is really the one the user needs. In this case, people may have to click on some of the entries with the hopes that the topic of interest will be within first few documents, or spend literally several hours navigating each of them. This is because these general-purpose search engines create indices for diverse topics and naive queries submitted by user often finds matches in many irrelevant pages

[61]. Generally, such search engines look for every occurrence of the word which was typed into the search interface. Upon finding them, it lists each and every document containing that word. However, the topic may only be mentioned within some of the

documents with no information of real value. Of course, people can get more relevant results if they provide more specific and accurate information to the search engines, but it will require the users to have much experience and sometimes the users are not sure how to narrow the search queries. Actually 70% web search users only use only one keyword, people are not accustomed to use sophisticated search functions like Boolean queries [62].

In some cases, when the topic related documents use certain words or phrases within the text, and although the users typed in synonymous terms, they cannot be matched by the search engines, and the worse case is that the keyword is simply misspelled.

Another issue with the general-purpose search engines is how to keep the web pages up- to-date. Since the Web is huge and web pages are updated frequently, the search engines have to refresh their indices periodically by crawling the Web and downloading web pages, which is extremely resource and time consuming. It is very difficult for a single search engine to cover the entirety of the Web and keep its index up-to-date at the same time.

Perhaps more important, the page ranking technologies with which search engines generate the order to list the results are operated by the commercial companies who own these general-purpose search engines, e.g. Google. Although Google takes a "trust us" approach to search and they will not skew their PageRank formula to favor certain sites, but users have no way of knowing for sure.

Finally, since the general-purpose search engines treat every webpage the same, many structured webpages lose semantic meaning after being indexed by the searching engines.

For example, the webpages on most online communities or social networking sites are actually meaningful with topics, posters, post time, post content etc. In addition, the general search engines have no way to monitor contents on the Internet beyond HTTP protocol.

To address the problems above, it's highly desirable to develop an integrated, topic/application-centered knowledge portal called Application-Specific Knowledge

Engines (ASKE) which supports more effective information retrieval and analysis as well as collaboration and communication among users. For example, if a user who is interested in the documents related to "sensors" in the "Intelligent Transportation

Systems" topic enters the single keyword "sensors", such knowledge portals are less likely to return irrelevant pages.

In this chapter, we demonstrate the concepts of ASKE and propose a feasible approach in many other different applications. The remainder of this chapter is organized as follows:

Section 3.1 presents the idea of knowledge portals and various applications of domain- specific search engines. We also review several approaches to building domain-specific

Web search engines. We present an overview of the system architecture of the ASKE framework in Section 3.2. Thereafter, we explain the core components of the framework in Section 3.3, and 3.4. The conclusions are given in Section 3.5.

3.1 Knowledge Portals and Applications

Currently the problem we are facing is not how to acquire information but that we already have far too much information - we are overloaded. It's infeasible to cure the problem by providing access to more information, or even by improving our efficiency in generating information. One possible way is to produce, acquire, transmit, and manage knowledge.

In November of 1998, a new concept called "Enterprise Information Portals (EIPs)" was presented by Christopher Shilakes and Julie Tylman [63]. They defined EIPs as

"applications that enable companies to unlock internally and externally stored information, and provide users a single gateway to personalized information needed to make informed business decisions". EIPs consolidate diverse systems including Content

Management Systems, Business Intelligence, Data Warehouse/Data Mart, Data

Management, and other data external to these applications into a single system which can

"share, manage and maintain information from one central user interface". As the key component of EIPs, the Content Management Systems process, filter and refine unstructured or semi-structured information in diverse documents and formats and store it in a data repository.

Based on the ideas of EIPs, Joseph M. Firestone presented "Enterprise Knowledge

Portals (EKPs)" [64] which is a type of EIP. An EKP is an EIP which is goal-direct toward knowledge production, knowledge acquisition, knowledge transmission, and knowledge management focused on enterprise business processes; and focuses upon, provides, produces, and manages information about the validity of the information it

supplies. Simply, Knowledge Portals provide information about the topics of interests effectively and efficiently, and also supply users with meta-information about what information users can rely on for generating new knowledge. Actually this concept is not restricted in enterprise applications only. Knowledge portals can be applied in various domains, e.g. in scientific research fields, knowledge portals may correlate the information in this domain with perfect accuracy and present it in a manageable form.

Compared the features of knowledge portals and search engines, we can find search engines provide a feasible and flexible framework to implement knowledge portals of diverse topics.

In recent years, some such domain specific search engines have been built. In scientific research domains, CiteSeer [65] is a scientific literature digital library that aims to improve the dissemination and feedback of scientific literature, and to provide improvements in functionality, usability, availability, cost, comprehensiveness, efficiency, and timeliness; ResearchIndex [66] is constructed for the searching of computer science papers; Cora [67] is also a search engine for computer science research papers; NanoPort

[68] is a comprehensive Web portal to serve the researchers, scientists, and practitioners in the nanoscale science and engineering (NSE) domain; AGROA [69] searches for software components and builds a database classified by component; even there is specialized search engine called Deadliner [70] which is used to find conference deadlines. There are also some specialized search engines that are related to people's daily life, for example: Camp Search (www.campsearch.com) allows the user to search for summer camps for children and adults; Movie Review Query Engine

(www.mrqe.com) allows the user to search for reviews of movies; Travel-Finder

(www.trave-finder.com) allows the user to find travel information with activity, category and location; Newstracker (nt.excite.com) and Moreover (www.moreover.com) are good websites for the latest news; FlipDog (www.flipdog.com) searches job postings at various

IT company Web sites, and builds an up-to-date and powerfully searchable index of job advertisements; Crafts Search (www.bella-decor.com) helps the user search web pages about crafts and provides search capabilities over classified ads and auctions of crafts;

CMedPort [71] is a cross-regional Chinese medical Web portal to provide access to

Chinese medical information over the Internet.

Observing the building processes of these specialized search engines, we find that the most popular approach to building domain-specific Web search engines is to only collect and index the relevant documents available on the Web. This approach is composed of the following main processing stages:

• Collection of domain-specific web pages.

• Information Extraction from the domain-specific web pages, involving two sub-

stages: parsing and indexing.

• Data repository to store the parsing and indexing information into a common

database (MS SQL Server, MySQL etc.).

The other good approach is to reuse the existing indices of those general-purpose search engines. Compared to the approach above, it requires less human efforts to build and maintain the word/phrase indices. This approach is a kind of meta-search engines [72]:

the specialized search engine forwards the user's query to one or more general-purpose search engine and catches the returned results. To guarantee that the documents returned to the user are relevant to the topic of interest, two kinds of middleware (or agents) can be implemented during the meta-search process. First method is that the domain-specific search engine passes the original query to large-scale search engines, and eliminates the irrelevant documents from the returned ones via a domain-specific filter [73]. The drawback of this "filtering model" is that it downloads many irrelevant web pages and spends much time to filter them out which consequently results the inefficient performance. The other "keyword spice model" [61] does not filter documents returned by a general-purpose search engine. Instead, it extends the user's query with a domain- specific Boolean expression (keyword spice) that helps narrow the original query better and forwards the modified query to the general-purpose search engines. This model is more efficient than the first one, but the relevance of the returned documents cannot be guaranteed since it relies much on the quality of keyword spice. And when the domain

(category) is broader, more human expertise and efforts are required.

To promote the relevance of search results and the efficiency of building the specialized knowledge portals, we propose to establish a new type of structured framework for collecting and analyzing information in application specific domains, called application specific knowledge engines (ASKE). The basic idea is to create a series of spider agents which can collect data from heterogeneous sources, build up semantic data repositories, and use a Knowledge Configuration File (KCF) to specify topics, keywords, searching sequences and schedules for query processing. The key features of the proposed ASKE

are user specific, application specific, and domain specific and the process of ASKE is report-motivated, semantic-based, and time-driven, instead of question-motivated and query-driven, as in traditional web searching processes [74]. The important task of ASKE is to offer both end users and applications a seamless access to knowledge contained in heterogeneous data sources.

3.2 An Overview of the Framework for Application Specific Knowledge Engines

Figure 3.1 shows the framework and components of an application specific knowledge engine, which contains two main phases: construction of Data Repositories and searching by KCF with result presentation.

Phase 1 construction of Data Repositories: this phase focuses on how to build a high- quality data collection that is comprehensive and relevant, including three main sub- phasess: data collection, data preparation, and data silo. In data collection phase, we use

Resource Identifier to locate data resources relevant to the application, Spider Agents which can deal with different types of data to collect all data sources, Content Filter to exclude noise and garbage data in the data collection; in data preparation phase, we use

Data Classifier to categorize collected data based on different file types, apply Parser and

Indexer to the data collection to build up indices, lexicon library, and searchable databases; in the final sub-phase, we create semantic data repositories based on texture documents and structured databases based on the data we collect in the last sub-phase via

Ontology Developer and Metadata Extractor.

Phase 2 searching by KCF with result presentation : with high quality semantic data collection, the components in this phase try to help users to locate information accurately and easily; search is conducted by KCF or in semantic ways; search results are presented to users with different components including list, keyword suggestion, summarization, categorization, and visualization.

3.3 Construction of Data Repositories

The construction phase contains mainly three sub stages: data collection, data preparation, and data silo.

3.3.1 Data Collection

In building an application specific knowledge engine, high-quality collections will be main factor that determines its usefulness and efficiency. ASKE should contain as many as relevant, high quality documents and as few irrelevant, low-quality documents as possible. To address this need, there are two general approaches to collecting relevant

Web documents: manual selection and automatic Web crawling.

Figure 3.1 The framework and components of an Application Specific Knowledge Engine 1. Current popular approaches to collect documents

For example, In the National Library of Australia's PANDORA (Preserving and

Accessing Networked Documentary of Australia) project (http://pandora.nla.gov.au/), all

Websites were reviewed and manually selected to make sure that only relevant sites were archived. Also ACT (Anti-Terrorism Coalition) applies a decentralized mechanism to identify terrorist websites, e-groups and forums manually by agents and users spreading all over the world, and has compiled a database of terrorist and pro-terrorist websites and e-groups (http://atcoalition.showsit.info/). Through manual selection, genuine relevant information could be collected.

Alternatively, in the National Library of Sweden's Kulturarw project

(http://www.kb.se/kw3/ENG/Default.htm), a Web crawler (also referred to as Internet

Spiders or robots) was used to automatically download Websites with .se country domain names or from Web servers known to be located in Sweden. Web documents are spidered efficiently via automatic Web crawling. In this approach, some algorithms have been developed to guide Web crawlers to locate Web pages and predict whether a URL is likely to be the relevant resource, e.g. HITS [75], PageRank [76], etc.

However, neither of the above two methods could be directly applied here. The manual selection method is not efficient and often results in low coverage. The automatic Web crawling method often introduces noisy into the collections, resulting in low precision.

These two methods offer few efforts to trace and monitor websites’ transitions and guarantee update-to-date high quality data collections.

Figure 3.2 The structure of Resource Identifer 2. Resource Identifier

We propose to use a recursive collection building procedure which combines both manual selection and automatic Web crawling methods. In this method, a limited number of seed URLs are first identified through careful and systematic manual selection. These

URLs are then captured using automatic Web crawlers. Favorite link pages (a specific type of Web pages where the Web masters list the URLs of some other Websites that they recommend.) are extracted from the captured Websites. Manual selection is performed again on the out-going URLs listed in those favorite link pages which are very likely to be URLs pointing to other domain or application related Websites. Then, we repeat the whole process on the newly identified seed URLs to build the collection.

This procedure is conducted by the Resource Identifier . Figure 3.2 shows the structure of Resource Identifier. It is responsible for locating the domain relevant resources, e.g.

the sites that are known to provide specific contents, web directories that are dedicate to gathering related sites and web pages in this application. It extracts the favorite links of those well-known sites which are believed to be related to the topics of interest. Also a lexicon reflecting the key concepts and terms as well as their usage in the domain is built by domain experts. It is apparently helpful to collect as many specific terms as possible in the lexicon. The Resource Identifier sends queries that are created based on the lexicon to multiple general-purpose search engines and retrieves the top 100 or 200 results as the part of resources in our content collection. These top results are usually of high-quality and much diversity, thus greatly improve the quality of the collection.

In the mean time, users can configure those favorite sites as the seed URLs in KCF so that the data collection is specialized for their own interests. In addition, the resulting resources are classified into different categories (e.g. News, Journal Papers, and Forums etc.) for generating reports in the future.

3. Spider Agents

After the resources of high quality have been identified, the document collecting is done automatically by Spider Agents . A Spider [77] is a program that automatically fetches

Web documents. It is called a spider because it crawls over the Web, and another term for these programs is "crawler". Because most Web pages contain links to other pages, a spider can start almost anywhere. In our approach, to make sure that fetched pages are really relevant to the domain, the spider is limited within particular hyperlinked resources from the Resource Identifier. To fetch more pages in shorter time, many spiders work in

parallel. Compared to large search engines, our data collection is small, thus generally we only start 10-20 spiders to crawl the Web.

There are several issues required to be solved in the design and implementation of spider agents, including what to spider, how to spider, and how often to spider (frequency). The

Resource Identifier resolves the first issue. It provides spider agents resources (URLs, etc.) to collect data. As the relevance of the resources to the domain and user expectation which is specified in KCF varies, spider agents cannot treat all the resources the same, e.g. some resources have to be monitored more frequently to get real time data. In the ASKE framework, we borrow the idea from physical sensors to construct Sensor Network to build up spider agents.

• Social Sensor Network

Sensors are popular in people’s everyday lives. They can provide information about a car’s condition, can enable smart buildings, can locate friends’ locations, and are being used in a various mobile applications. Generally, sensors provide information about various aspects of the real world. Similarly, “sensors” can also be used on the virtual world – the Internet to “sense” the Web data. As information appears and spreads on the

Web, we build up various “social sensor networks” (see Figure 3.3) in different circles based on the categories or organizations of resources and the relevance among them, actively monitoring the resources by collecting data in real time.

The Internet is a huge dynamic system. Given a URL, the content in this URL may change frequently as time goes by, such as responses (comments, replies etc.) are added

to the Web page. In P2P applications, new ed2k links or torrents for a given title can appear or disappear at any time, and users that download the corresponding copyrighted contents are more inconstant. By periodically monitoring and scanning resources, we may discover if new data on the Internet is interested by users in real time. The list of monitored resources and scan intervals can be customized by users via KCF.

Category/ Organization 1 Category/ Organization 2

Category/ Organization 8 Seed 1 Seed 8 Category/ Seed 2 Organization 3

Seed 7 Domain/ Application Seed 3

Seed 8 Category/ Seed 4 Organization 7 Seed 5 Category/ Organization 4

Category/ Organization 6

Category/ Organization 5

Figure 3.3 The architecture of social sensor networks

In social sensor networks, scan intervals can also be changed adaptively in the framework by calculating activity levels of resources. The closer a resource is to the center, the sooner the resource is scanned. For example, initially we monitor a torrent file “Avatar

2009 DVDScr H264 AAC-SecretMyth (Kingdom-Release)” every one hour. In the first hour, we found 34 peers downloading this torrent, and in the second hour 175 peers downloading this torrent which are 5 times of the first hour. In this case, the spider agents would reduce the period from 1 hour to 12 minutes as this torrent becomes a “hot” resource.

Although Web spider technology is kind of mature to handle most of unstructured web pages, as the new technologies and applications are applied in the Internet, the traditional

Web spiders cannot handle a lot of new applications, such as Ajax [78], online communities, and P2P applications.

• AJAX Applications

The traditional spidering process is as the following: connecting to a server, pulling down the HTML document, parsing the document for anchor links to other HTTP URLs and repeating the same process on all of the discovered URLs. Each URL represents a different state of the traditional web site. AJAX (shorthand for asynchronous JavaScript and XML) is a group of interrelated web development and communication techniques running on clients to give users the same interactive experience as desktop applications.

With Ajax, web applications can retrieve data from the server asynchronously in the background and update the part of the existing web pages without refreshing or reloading

the whole page. Ajax techniques make web applications feel like local applications with rich interactivity and dynamics.

In a Web page with AJAX enabled, the page is loaded from the web server in the beginning. When the user starts to interact with the Web page, like clicking buttons, selecting options etc, some of the page content is changed in the HTML document, which is dynamically inserted by Javascript during the interactions process. Not only form elements can trigger Javascript events, but also anchor links which usually point to other

URLs. After the Web page is fully loaded, the series of Javascript events that are triggered define the states of the application. For AJAX applications, the traditional spider can only reach or fetch a small fraction of the content and is unable to index any of the application's state information and dynamic contents.

Obviously, traditional web spiders would not work with AJAX applications. To solve this issue, new spiders that can understand not only how to traverse among links and parse

HTML format, but also the structure of the document as well as the Javascript that manipulates it have to be developed. To be able to research the complete states of an application no matter how deep the state is, the new spiders also need to be able to identify and trigger events within the document to reach each possible state and simulate the paths that might be taken by a real user.

As we discuss above, traditional spiders run in a “protocol-driven” way, which does not work they meet an AJAX enabled page. This is because all target resources are generated by JavaScript codes dynamically and are embedded in the Document Object Model

(DOM) context. It is important to both understand and trigger this DOM-based activity.

In the process, new spiders have to implemented in another way called "event-driven"

[79] . It has the following three key components:

o Javascript analysis and interpretation

o DOM event handling and dispatching

o Dynamic DOM content extraction

To develop a complete new Javascript interpretation engine would be a very tough and time consuming job. By using a modern browser such as Firefox or Internet Explorer (IE) as the underlying platform, we implemented our AJAX-enabled, event-driven spiders in an efficient way. There are a couple of similar tools available to utilize the existing browsers, such as Watir [80] and Crowbar [81]. These tools allow us to control Firefox or

IE from our own codes, thus to extract data after any Javascript event is triggered.

Watir, pronounced water, is a library that enables automating web browsers using Ruby.

It supports most of current modern browers, such as Firefox and Safari. The Watir API allows users to launch a browser process and then directly extract and click on anchor links from Ruby application.

Crowbar is another interesting tool which uses a headless version of Firefox to render and parse web content. By providing a web server interface which wraps the browser, it allows us send simple GET or POST requests from any language or even simple command line tools such as curl and wget, and then fetch and parse the results as needed just as traditional spiders do.

We chose Crowbar instead of Watir after careful comparison. Crowbar is language independent and simple to integrate into an existing mature traditional crawler design to extract page information that would only be available after a page has completely loaded.

Watir, on the other hand, provides deeper, interactive, and direct access to the browser, thus is easier for users to control and trigger Javascript events. The main issue is that we have to develop a brand new spider with more complicated logic to discover states and additional Ruby exploration.

• Online Forums

Most online forums look like same as common Web pages. But besides the regular spidering process, we have to consider some other issues when we design the spider agents specifically for online forums.

1) Apply for forum membership. Many online communities require membership to access.

We need to create a common user name and then send the application request to forum masters. Once the application is approved, we use the user name and password to access the forums. In some cases, the forum masters could be very selective. It may take a couple of rounds of emails to obtain access privilege.

2) Handle membership access in spidering. We manually access the forums for the first time using the authorized user name and password. The forums store our access information in permanent or temporary cookies on our local computers. When spidering the forums, we direct our spider program to use the cookies with authorized membership information to access the forum contents. We follow Robot Exclusion Protocol. It is a

method that allows Web site administrators to indicate to visiting robots which parts of their site should not be visited.

3) Handle different forum software. Online communities are implemented using different forum software packages with different Web application languages. To download these forums with different packages, the spider program needs to know the pattern of URLs created by such software. Thus, once the forum software packages are identified, URL templates need to be created. The spider program then uses the templates to generate parameters required by corresponding forum software packages. Table 3.1 shows forum software packages used in most online communities and their corresponding application languages.

Software Packages Application Languages Software Packages Application Languages

Discuz! PHP PHPWind PHP

Crosstar PHP phpBB PHP

DCForum PHP rafia PHP

ezboard CGI vBulletin PHP

IM PHP WebRing CGI

Invision Power Board PHP WebWiz ASP

newbb PHP YaBB PHP

Table 3.1 Popular Forum Software Packages 4) Identify patterns of forum messages. Besides the different URL template needed for each forum package, different syntax and formats in forum messages also need to be captured for different forums software packages. In order to correctly extract important information such as thread titles, authors, and post dates, the spider program need to

recognize these important fields from the returned messages. We create separate parsers for each forum software listed in Table 3.1 to extract the most essential information from forums of different formats. For example, for messages from vBulletin-based forums, our vBulletin parser is able to extract the message titles by seeking the title meta tags (“” and “”) in the message body.

Protocol Type of Server

http:// HTTP server

https:// Secured HTTP server

mms:// Microsoft Media Player stream server

mmst:// Microsoft Media Play TCP stream server

pnm:// Older version of Real Player stream server

rtsp:// Real Player stream server

ftp:// FTP server

Table 3.2 Different Types of Download Servers 5) Handle external links and local attachments: In addition to the textual contents of forum messages, multimedia materials (e.g., images, video/audio clips, etc.) posted by forum participants in their messages are also very important. This material can be uploaded to the forum server as attachments or can be hosted on another server with their

URLs posted in the forum message as external links. We set up our spider program to not only download Web pages, but also download textual documents (e.g., plain text files,

MS Word files, PDF files, etc.), multimedia documents (e.g., images, animations, video/audio clips, etc.), archive documents (e.g., ZIP files, RAR files, etc.), and other non-standard files (files with extension names not recognizable by Windows operating

system). The external materials are hosted on different types of servers that require different download methods. The spider program monitors the protocol sections of the external URLs to decide which download method to use. Table 4 summarizes the types of download servers that the spider program recognizes.

6) Handle Multiple Views of the Same Content: Forums sometimes support views of the same content. For example, in vBulletin-based forums, a thread can be displayed in three views: linear view, threaded view, and hybrid display. Each view is spidered as a unique document although they contain the some content. These redundant documents need to be filtered based on the URL patterns to keep the collection concise. Furthermore, in a forum, there are pages that do not contain meaningful information, for example, pages where users start a new thread or pages where users post replies to an existing thread.

These non-meaningful pages also need to be filtered out.

7) Prevent Spiders from Vicious Links: Some forums may contain hyperlinks that trap a spider program in a loop or dynamically generate infinite number of new links (e.g. calendars, forum internal search engines, etc.). If the spidering process does not finish in a reasonable time limit or after a reasonable number of documents are downloaded, we need to examine the spidering log and identify the vicious links. Identified vicious links are excluded in future spidering process.

8) Prevent Spiders from Being Blocked: Forums may block an IP address if too many requests are sent from the IP address in a short period of time. We need to set a random time delay between hits and make the spider program mimic human browsing behaviors.

Some forums only allow specific types of Web browsers (e.g., Microsoft Internet

Explorer 8.0, Firefox 3.6, etc.) to access their contents. We need to set up the spider program such that it mimics a certain Web browser to access the target forums.

9) Update Forum Collection: Forum contents are constantly being updated. The spider program needs to revisit the target forums periodically to download new threads and posts.

In summary, the proposed forum collection approach takes advantage of both human experts’ knowledge and automatic Web spidering techniques. It reduces human intervention and increases the efficiency of forum collection and monitoring process.

• P2P Applications

To collect data about how people use P2P applications to download resources and monitor their behaviors, the best way is to be part of P2P network. We create multiple spider agents implementing different P2P protocols (e.g., BitTorrent, eDonkey, and

Gnutella). Since spider agents would not really download/upload contents, they act as hidden “sentinels” with pretty light weight mimicking different roles: regular or “neutral” peers, malicious peers, and seed servers.

As ‘neural” peers, the agents mimic the behavior of P2P peers by implementing the same discovery and download protocols, exhibit similar download speeds, arrival and departure rates as the regular clients, and in the mean time collecting their neighbors detail information, e.g., IP address, shared files (name, size), .

As “malicious” peers, the agents mimic the behavior of malicious peers by sending out probes to their neighbors at the same rate as other malicious peers. The agents actively search, study and run the software that malicious users use, and also use two-way communication packets to verify the malicious peers’ IPs.

Finally, the agents serve in the capacity of BitTorrent Tracker servers and eDonkey index servers, and share real copyrighted contents to figure out who download them from us.

Since the spider agents are light-weight, using very little machine resources (CPU time, memory etc.), we can spawn hundreds of instances on one server with high band-width network connection. By assigning servers over USA in different locations, we can expect to monitor P2P activities in USA in real time.

4. Content Filters

Although the Resource Identifier tries to only get the relevant information, some noise inevitably exists in the collection. And sometimes the noise is brought by the spiders that catch some irrelevant information. The common used filtering techniques include:

• Domain experts manually determine the relevance of each Web page.

• In the simplest automatic way, the relevance of a Web page can be determined

by the occurrences of particular keywords (e.g., health care bill ) [82].

• TF*IDF (term frequency * inverse document frequency) is calculated based on

domain-expert created lexicon. Web pages are compared with a set of relevant

documents, and those with a similarity score above a certain threshold are

considered relevant [83].

• Text classification techniques, such as Naive Bayesian classifier [84] [67] and

Support Vector Machine (SVM) [85].

Our approach is to calculate the sum of two values of a given webpage via the terms and combinations of the terms in the domain lexicon:

o TFIDF(p) = Sum of TF*IDF of the terms in page p found in domain lexicon

o Title(p) = Number of terms in the title of page p found in domain lexicon

The relevance of each document is calculated and ranked. The quality of the collection can be guaranteed by a relevance threshold which is calculated from a set of manual identified relevant documents.

3.3.2 Data Preparation

After collecting Web documents and different kinds of data, we need to process the collection and make them accessible or searchable.

1. Data Classifier

Generally, the collection contains diverse types of documents, e.g. web pages, PDF and

Word files, images etc. In some domains, especially scientific domains, people would like to find information in different types separately. The "Data Classifier" classifies the documents based on their file types.

2. Parser

The Parser converts different types of files into the plain text format, so that later on the

Indexer can process them. For webpages, the Parser recognizes the html tags and words.

The html tags, style sheets, and scripts are stripped off. For other types of documents, like

WORD or pdf, the parser utilizes tools (e.g., pdf2txt, ABC Amber Text Converter, etc.).

For multimedia objects, we mainly use the texts around them.

3. Indexer

The Indexer is responsible for tokenizing the Web data into words, and storing the relationships between words and documents. Various information techniques are applied in the Indexer, e.g. stop words and fuzzy indexing. To support the fuzzy searches (for example, a search for "running" also finds "run" and "runs"), several stemming algorithms are used: the Porter Stemming algorithm [86], Soundex, Metaphone and

Double Metaphone algorithm [87]. It is a good practice to create both a normal (exact) index and a fuzzy index and allow the search interface select which index to use. The indexer doesn’t need to process any P2P data, since they are already structured. Finally the resulting searchable indices are stored into a database for further document retrieval and analysis.

3.3.3 Data Silo

After the first 2 sub stages ‘data collection’ and ‘data preparation’, we now have the collection of all available data sources such as semi-structured textual documents,

structured data in the form of databases. In this stage, we need to extract high quality metadata from heterogeneous sources in an automated manner to build up the semantic data repository, which is part of “Data Silo” combined with previous available data collections.

The Semantic Web [88] is the vision of the next generation of the World Wide Web, in which information is given well-defined meaning and thus becomes understandable and consumable not only for humans, but for machines as well. Currently there are some semantic web portals focused on certain domains that have been developed. In these web portals semantic mark-ups are gathered, stored and accessed, for example MindSwap [89],

Knowledge Web [90], and MuseumFinland [91]. Semantics are added to the contents and services provided by these semantic web portals extending the traditional idea of web portals. MindSwap and Knowledge Web are examples of research projects based semantic web portals. They describe and present relevant domain knowledge based on back-end semantic data repositories. As the first semantic web portal, MuseumFinland aggregates heterogeneous museum collections. The underlying metadata is extracted from distributed databases by mapping database schemas to the shared museum ontologies. The portal provides a view-based multi-facet searching function and a simple keyword searching function. Via multi-facet searching, users can filter results using museum category hierarchies. The keyword searching function, in another way, matches the keyword with the available categories and then uses the category matches to filter results.

Two components in this stage have been developed: Ontology Developer and Metadata

Extractor.

1. Ontology Developer

An ontology is a formal specification of a shared conceptualization [92]. Ontologies capture the structure of the domain, i.e. conceptualization, including the model of the domain with possible restrictions. The conceptualization describes knowledge about the domain, not about the particular state of affairs in the domain. In other words, the conceptualization is not changing, or is changing very rarely. Ontology is then specification of this conceptualization - the conceptualization is specified by using particular modeling language and particular terms.

There are a lot of existing ontologies in our daily life, for example:

• Taxonomies on the Web: DMOZ (Open Directory Project), Yahoo! and Google

categories

• Catalogs for on-line shopping: Bing Stores, Amazon product catalog

• Domain-specific standard terminology: Unified Medical Language System (UMLS),

UNSPSC - terminology for products and services

Usually a domain specific ontology is developed by domain experts. Figure 3.4 shows the ontology development process, which includes 7 steps. These steps may repeat during the whole process.

Figure 3.4 Ontology-development process

• Determine domain and scope: this step tries to answer questions, like what the

domain that the ontology will describe and cover is, how we are going to use the

ontology, for what types of questions the ontology could answer, and who will

use and maintain the ontology, etc.

• Consider reuse: are there any available ontologies that have been validated

through use in other different applications?

• Enumerate important terms: find out the terms and the properties of these terms,

and how we are going to describe these terms.

• Define classes and the class hierarchy: A class can be defined as a concept in the

domain or a collection of elements with similar properties, and classes usually

constitute a taxonomic hierarchy.

• Define properties of classes: include “intrinsic” properties, “extrinsic” properties,

relations to other objects etc.

• Define constraints: property constraints that describe or limit the set of possible

values for a slot, and constraints that defined for class inheritance.

• Create instances: the class becomes a direct type of the instance, and assign

property values for the instance which should conform to the constraints.

The Web Ontology Language (OWL) which is a family of knowledge representation languages including OWL DL, OWL Lite, and RDF schema, is used this component to author ontologies. In this component, domain experts follow the process we describe above to develop ontologies using RDF/XML syntax.

Figure 3.5 The architecture of Metadata Extractor Developing a new ontology for a specific domain or application is one way. The alternative way is to find out existing validated ontology for this given domain. There are a lot of available sources for ontologies online, e.g., semanticweb.org, Swoogle

(http://swoogle.umbc.edu/).

2. Metadata Extractor

The metadata extractor is responsible for the extraction of high quality semantic data from the source data. How to ensure the quality of the extracted data becomes a challenged issue. In the metadata extractor, we design several ways to extract high quality data. First, it combines the context in which an entity is mentioned in order to determine its type and thus reduce ambiguities. Second, a verification engine is developed in the metadata extractor, and it checks the validity of any new derived metadata by comparing them against a trusted domain knowledge collection and the related information on the Web. Finally, we keep the metadata extractor automatically running whenever new data is pumped in, thus ensuring that the semantic metadata is always up to date. Figure 3.5 shows the architecture of the metadata extractor. It contains two key components: an automatic and adaptive metadata extraction tool used to marks- up textual sources; a semantic transformation engine, which convert and format data from original representations into new formats based on the specified domain ontology.

• Metadata Extraction

To address the issue of adaptive information extraction, we use Arizona Noun Phraser

[93], a named entity recognition (NER) tool that provides an adaptive service. The AZ

Noun Phraser is made up of three major components, a tokenizer, a part-of-speech tagger, and a phrase generation tool. It uses textual documents or files as input and generates a list of the named entities mentioned in that document. It relies much on domain ontologies and a repository of lexicon entries to process heterogeneous documents.

For the purpose of converting the extracted data to the specified domain ontology, a semantic transformation engine has been developed. It generates semantic relations along with semantic data entries among the named entities from Noun Phraser, and the specification of domain specific knowledge (i.e. lexicons, which are used in the

Resource Identifier ). The lexicons will be later used by the verification process also.

The inputs for this component are structured sources from Noun Phraser and transformation instructions.

• Metadata Verification

The goal of the verification engine is to check that each entity has been extracted correctly by the extraction components to ensure the high quality of extracted data. The verification process consists of three complex steps which involve several semantic web tools and some resources.

Step1: Checking the internal lexicon library. The lexicon library maintains domain specific lexicons, such as abbreviations, and records the mappings between strings and instance names. The verification engine considers any abbreviation is matched to the corresponding entity.

Step2: Querying the semantic web data repository. This step queries the existing semantic web data which is acquired before and should be correct. The querying process is conducted by a number of string matching algorithms (such as Rabin-Karp string search algorithm, Knuth-Morris-Pratt algorithm, Boyer–Moore string search algorithm) to deal with minor errors and typos in the data. If there is a single match, which means

the current data entry exists in the semantic web data repository already, the verification process ends immediately. However, there could be more than one match. We have to utilize contextual information to address the ambiguities.

Step3: Investigating external resources. When the second step cannot find a match for this entity, external resources such as the Web or existing semantic resources are applied to identify whether the entity is erroneous, which should be removed or corrected. Here we make use of PANKOW classification service and WordNet [94], to determine the proper classification of the entity. If the entity is not classified correctly, we have to compare other major concepts of the domain ontology with the Web-endorsed type to find an appropriate classification for the entity in the domain ontology. Otherwise, we can just create a new instance and add it to the repository directly.

3.4 Searching by KCF with Result Presentation

When a collection is ready, the next phase, consisting KCF Processing, Keyword Search,

Semantic Search, and Result Presentation, is conducted to support searching and analysis of the results.

3.4.1 KCF Processing

Quite often, it is impossible to create an application that works for all users. Each user may have his or her own preferences or requirements and the application needs to be able to adjust accordingly at runtime to satisfy the user. One of the easier approaches is to use

configuration files to remember the user's preferences. Knowledge Configuration File uses XML to describe users’ preferences and customize the system’s behaviors.

KCF includes mainly 4 sections: user profile, favorite resources, favorite topics/categories, and queries.

1. User Profile

This section stores a user’s personal information, such as name, email, and specialty etc.

2. Favorite Resources

In this section, users provide favorite resources that are monitored by the system frequently. The resources could be a URL, an ED2K link, a torrent file link, or a title for copyrighted contents. Each resources have three attributes: topic, period, and type. Topic attribute states the topic or category this resource belongs to; Period attribute customizes the frequency the Spider agents monitor the resource; and Type attribute points out what kind of the resource is, such as URL, ED2K link etc.

…

3. Favorite Topics/Categories

Favorite Topics/Categories section stores the topics or categories that the user is interested in. This helps the keyword search and semantic search only find related information in these predefined categories. Each topic is described with name, description, and a series of keywords. If a topic is a sub-topic of another topic, it also has a parent_id attribute to describe which parent topic it belongs to.

…

4. Queries

In this section, users can store searching sequences and schedules, which help minimize the efforts to create complex search queries manually in one search.

0-59

0-23

1-31

1-12

0-7

…

The “schedule” section follows the configuration rules in common CRON jobs. It defines how the searches are performed periodically. The keyword sections define the keywords and the sequences that will be searched in the system. The Type attribute in keyword

sections defines how the keywords will be searched, such as “phrase searching”,

“Boolean AND searching”, “Boolean OR searching”.

The KCF Processing module mainly performs the following tasks:

• Parses the preset knowledge configuration file which is in XML format, extracts

topics, keywords, searching sequences and schedules;

• Following the searching sequences and schedules, generates corresponding queries;

• Parses each query into several words with Boolean relationships or phase searching;

• Forwards the each query to the data repository to find search results recursively;

• Synthesizes the results from each query based on the relationship configured in KCF.

3.4.2 Semantic Search

Keyword search is currently most popular way to identify resources in search engines, as it provides a simple way to query data. Compared to keyword search which is based on keyword comparison, semantic search provides better results and performance for keyword searching by using the underlying ontologies and metadata collection built in semantic web portals. Most of current semantic search technologies [95-97] mainly focus on enabling semantic entities search, which totally neglect the semantic relations among entities which are generated by manual annotation and automatic/semi-automatic extraction.

Here we use the keywords and topic defined in KCF as input to do the semantic search.

The topic is used to limit the search procedure in some specific domain. The procedure will look like as the following:

1. Understand the input

A keyword may have different meanings in different context. It can be general concepts, or instances entities, or even literals (e.g., strings) of certain instances. This step employs string matching algorithms to match the keyword against literals, ontology concepts, and instances step by step and finds out the keywords falls into which category: general concepts, instances entities, or literals. Then we can know which entity is matched and the similarity for the input keyword.

2. Get related instances

If the keyword is a general concept, the related instances are the instances of the matched concepts. If the keyword is an entity, the related instances are the matched instance.

3. Get search results

The search results are actually those instances that have direct or indirect relations with the matched instances. If the instances are associated directly in explicit triples, we call it a direct relation, while indirect relations are the ones which can be derived from the explicit triples. Relations can be associated together when they share a same instance, which is called Mediator. The more mediators exist in between, the weaker the derived relation is.

4. Rank search results

Search results are ranked based on string similarity, and semantic relation closeness. The string similarity is calculated in Step 1. The semantic relation closeness represents how close the semantic relations are between the instances in a search result and the matched instances. To quantify the semantic relation closeness, we count all the relations a search result has with all of the matched instances, for example, we give the direct relations the weight 1.0, and the derived ones 0.5, 0.25 etc.

3.4.3 Result Presentation

The Result Presentation module is the key role to generate reports and send them to users.

It contains five functional components: the Document List, the Keyword Suggestion, the

Summarization, the Aggregation and the Visualization. Consolidating these components, the system will present users an integrated, comprehensive, systematic and up-to-date report.

1. Document List

It will list all the results from various sources one by one from the high relevance to the low one, just as general-purpose search engines do. In reports, the address of each document, size, modified time, short description and cached location are presented to the users.

2. Keyword Suggestion

A lot of research work has been done in the area of keyword suggestion. There are mainly three different techniques for generating keywords: query log and advertiser log mining, proximity searches and meta-tag crawlers. The first technique is popularly used by search engines, as they have huge amount of query log data and advertiser log data.

By mining the logs, they can figure out what keywords are usually queried together with the initial keywords (we call co-occurrence relationship between the keywords). For example, Google's Adword Tool [56] mines advertisers' log to determine keywords they searched. Bartz [98] proposed a new method based on collaborative filtering which involves the clicked URLs as a factor, and uses the relationship between the URLs and the query terms to generate new keywords. Obviously, it’s possible that the suggested keywords are only the ones appear the logs most frequently, which can be unrelated to the initial keywords.

Most of the commercial tools use the second technique - proximity based methods - for generating keywords. The suggested keywords generated by search engines are used as the seed keywords. The tools then append words found in its proximity.

Keyword Suggestion component helps users to expand and optimize ineffective keywords by showing users related keywords that include the search terms or phases. As we have created semantic data repository, we implement keyword suggestion component via computation of semantic similarity by using a diffusion process on a graph defined by lexicons and co-occurrence information [99].

3. Document Summarization

Summarization can be performed on individual documents to provide summaries which are intended to give quick previews of Web documents [100] by reducing the size and complexity of the documents and offering concise representations of the documents [101,

102]. For example, for a long web document, via document summarization technique the key sentences are extracted, thus to allow users to make a decision whether a document is of interest without reading the document in detail based on the relevance of the document.

There are mainly two methods to text summarization, including text extraction and text abstraction. Text extraction, as the name suggests, extracts original sentences from a document to construct a summary. The summarization involves the following related techniques: sentence evaluation, segmentation or topic identification, and segment ranking and extraction. Text abstraction, in the other hand, summarizes a document using generated grammatical sentences. Obviously, a great deal of document processing and computation has to be involved. From the review, recent research trend of document summarization focuses on the text extraction method [102, 103].

The Summarization component is responsible for summarizing texts from the search results. It reads texts from the resulting Web documents, decides which sentences are important and which are not, and adjusts the summary ratio.

4. Document Categorization

Users are often frustrated when facing a huge amount of documents, so it is desirable to gain an overview of the result documents, explore different topics, and gain a general idea of a particular area of interest. Automatic document categorization [104, 105] is the

technique to help aggregate similar or related documents and present the resulting clusters to the user in an intuitive and sensible way [106].

In recent years, various document categorization algorithms have been developed [105,

107-112]. Document categorization is based on the Cluster Hypothesis “closely associated documents tend to be relevant to the same requests” [113]. These algorithms can be divided into two categories. The first category depends on traditional machine learning algorithms such as support vector machines, probabilistic classifiers, decision trees, rule sets, instance-based classifiers, etc. The second category contains specialized categorization algorithms for information retrieval, e.g. relevance feedback, linear classifiers, generalized instance set classifiers. Document categorization is usually implemented based on individual document attributes or inter-document similarities. In all algorithms in both categories, key phrases are identified by systematically segmenting and indexing documents. Arizona Noun Phraser [93] and Mutual Information [114] are the two tools to extract key phrase.

The Document Categorization component is used to provide users with a list of categories which are decided in the KCF so users can explore the results by categories as well as by ranking orders in the Document List

5. Document Visualization.

Data visualization is to facilitate the access of large volume of data and reduce information overload when a large number of search results are obtained. Card,

Mackinlay and Shneiderman defined data visualization as “The use of computer- supported, interactive, visual representations of abstract data to amplify cognition” [115].

Applied in online search, visualization techniques help users identify the most relevant documents from the entire set of retrieved documents. When the entire set of retrieved documents is returned, it is represented in the visualization format. The user can get an overview of the retrieved documents. Also the user can identify clusters of documents which have similar features, trends in documents from different websites, and relations among documents etc. Examples of such visualization techniques are the Jigsaw [116],

Geographic Information Systems (GIS) [117], Starfields [118] and Venn diagrams [119].

The Document Visualization component presents the results as a topic Self-Organizing

Maps (SOM) [120] that can be visualized as a 2-dementional neural network when the classification process is applied. This function component is mainly implemented by expanding the functions of 2-dementional information organizer for web documents [121] based on SOM. The SOM is a good approach that automatically arranges high dimensional statistical data so that similar documents are mapped close to each other.

3.5 Conclusions

In this chapter, we propose the concept of ASKE and our design of an ASKE framework for information retrieval and analysis. We discuss the details of components in the framework, including related research works, ideas behind each component, and implementation details.

CHAPTER 4

SEARCHING TERRORIST GROUPS ON THE INTERNET

Terrorism threats have a wide range that spans personal, organizational, and societal levels and have far-reaching economic, psychological, political, and social consequences

[122, 123]. Terrorists are using modern communication and information systems, especially the Web, to their own advantage. The Web has evolved into a global platform for people to use in disseminating and sharing ideas due to its easy access, anonymity, and international character; however, terrorists are utilizing the Web for their relocation, propaganda, recruitment, and communication purposes.

Nowadays, all active terrorist groups have established their presence on the Internet [124] via websites or online bulletin boards. In daily life, many terrorists, extremist groups, hate groups, and racial supremacy groups seize upon the worldwide practice of using the

Internet to improve communication and aid organization, allow members to coordinate quickly with large numbers of followers, and provide a platform for propaganda and even training manuals. The Web also allows terrorists to reach a wide audience of potential donors and recruits who may be located over a large geographic area. For example, the famous racial supremacy group, KKK, uses a Web site (http://www.americanknights.com) to distribute their ideas. The international terrorist group, Islamic Jihad, also uses a Web site (http://www.abrarway.com) to disseminate their ideologies, develop their strategic intelligence, and attract potential group members. In addition, terrorists are exploiting online public forums such as Yahoo’s eGroups, Usenet discussion forums, bulletin

boards, and chat rooms. We call this alternative side of the Web, which is used by terrorists, extremist groups, and their supporters, the Dark Web .

Since the Dark Web datasets are generated and/or used by terrorists and their supporters, they would be analyzed to enable better understanding and analysis of the terrorism phenomena from the “terrorists’ point of view”. These contents provide snapshots of terrorist activities, communications, ideologies, relationships, and evolutionary developments. However, traditional approaches to study terrorism groups are not applicable to collecting and analyzing Dark Web information due to lacking of a lack of advanced methodologies for data collection and mainly relying reliance on manual analysis [125], the traditional approaches to study terrorism groups are not applicable to collecting and analyzing Dark Web information.

There are several problems that are preventing effective and efficient discovery of Dark

Web intelligence. The first problem is mainly associated with information overload. The amount of data available on the Web is often overwhelming and unmanageable to the counterterrorism experts. Also, there are large and scattered volumes of terrorism-related data available from diverse sources available to analyze terrorist threats and system vulnerabilities [126], and counterterrorism experts are hindered from integrating these diverse sources and obtaining a comprehensive picture. There are currently no advanced or new methodologies to identify, model, and predict linkages among terrorists and their supporters. Another problem is that data posted on the Web are not persistent and may be misleading. The resources may suddenly emerge, frequently modify their formats, and

then swiftly disappear or, in many cases, seem to disappear by changing their URLs but retaining much of the same content [124]. Thus there is an urgent need to preserve these resources before they are forever lost. This last problem mainly involves language barriers faced by counterterrorism experts when dealing with the multilingual contents of terrorists’ emails from all over the world. Due to these problems, there is no general methodology for collecting and analyzing Dark Web information. Even if Dark Web datasets are collected, without analytical tools, raw data are still not useful for the experts to produce more research of genuine explanatory and predictive value. Thus, developing advanced techniques to support intelligent information archiving, searching and new approaches to analyze and map terrorism knowledge domains is an urgent and challenging problem.

In this chapter, we will describe how the ASKE framework is applied to this specific domain to develop an experimental Web-based counterterrorism knowledge portal, called the Dark Web Portal, to support the discovery and analysis of Dark Web information and provide an intelligent, reliable, interactive, and convenient interface with for the counterterrorism experts. The remainder of this chapter is organized as follows. Section

4.1 reviews existing terrorism research portals that provide terrorism-related information to experts and researchers. Section 4.2 presents the research questions. In Section 4.3, we report our experience implementing of the Dark Web Portal in details. Section 4.4 provides our concluding remarks and suggests the future research directions.

4.1 Literature Review

Organization Acquisition/Collectio Cataloging/Identificatio Storage Preservatio Access n n n

Archive 1. Internet 1996 -. Spidering to Metadata such as URLs, Server Ongoing Via Archive (IA) collect open access status, owner, data project HTML pages created, collection web site Research Center 2. Anti- 2003 -. Has 448 Metadata such as URLs, Database, Not Via Terrorism terrorist web sites & status, owner, data Server mentioned project Coalition egroups created, ISP, group web site (ACT) affiliated 3. Prism 2002 -. Limited # of Server Not Via (ICT, Israel) web sites. Project for mentioned project Research of Islamist “ “ web site Movements, Reuven Paz, Director 4. MEMRI 2003 -. Jihad & Not store None n/a Terrorism Studies “ “ content Project 5. Site 2003 -. Manual Server Not Via Institute collection, Rita Katz, “ “ mentioned project Director web site 6. Weimann 1998 -. Manual Database Ongoing Closed (Univ. Haifa, collection “ “ to public Israel) Vigilante Community 7. Internet 2001- . Spidering. Database, Not Via Haganah Has 100s links to web Server mentioned project sites. Confronting the “ “ web site Global Jihad Project 8. Simon 2004 – tracking 4000 Not Not Via Wiesenthal problematic websites, mentione mentioned Compac Center, and highlighting over d t Disks Snyder 200 of these sites Social Action Institute 9. 2001 - spidering Server Stores back Via Johathanrgal Has 60-70 sites. copies project t (geocities) Monitors sites that web site closed. Table 4.1 Current Approaches to Archive Terrorists’ Web Resources

In this section, we review current digital archiving for terrorists’ resources and existing terrorism research portals provided by specialized research centers or vigilante communities.

4.1.1 Digital Archiving for Terrorists’ Resources

Since the post September 11 war on terrorism has increased terrorist groups’ dependence on the Internet, researchers, journalists, archivists, and vigilantes (including hackers) are undertaking measures to ensure the continued but controlled availability of terrorists’

Websites for research and counterterrorism purposes. They are creating databases and

Websites for monitoring, collecting, classifying, preserving, and publicizing terrorists’

Websites. For example, the Jihad and Terrorism Studies Project by the Middle East

Media Research Institute (MEMRI) and the Project for Research of Islamist Movements

(PRISM) by the Interdisciplinary Center Herzilya, Israel, monitor Websites of militant

Islamic groups in their native languages and provide access to translated information and metadata about the groups’ Websites and forums.

Table 4.1 lists the organizations involved in trying to preserve terrorists’ web-based resources, which can be grouped into three categories: archive (1), research center (5), and vigilante community (3). It uses Hodge’s digital preservation life cycle [127] to summarize the activities of the organizations. The life cycle provides a framework for managing digital preservation and includes six stages: creation, acquisition/collection, cataloging/identification, storage, preservation, and access. Since we have limited knowledge in identifying who authors a terrorist’ Website, the creation stage will not be

included because of the clandestine aspect of terrorist activities and anonymous nature of

Internet. Table 4.1 starts with acquisition and collection development, the stage in which either the terrorist’s Website or information about the Website is acquired and collected.

Only three of the organizations as listed in Table 4.1, acquired terrorist’s’ Websites or forums. The others only provide access to terrorists’ groups URLs and selected metadata such as the name that the Website is registered under in the Whois directory, ISP, and date created. Most of these organizations store digital copies of terrorists’ Websites. In terms of preservation approaches, only Internet Archive describes how they preserve

Websites. Preservation of terrorists’ Websites and forums is at a nascent level.

4.1.2 Terrorism Research Portals

In analyzing terrorism phenomena, terrorism research portals provide services which help researchers locate, collect, and analyze Dark Web data. There are already numerous information portals provided by specialized research centers such as the Center for the

Study of Terrorism and Political Violence (CSTPV), located at St. Andrews University,

Scotland, and directed by noted terrorism researcher, Professor Paul Wilkinson and formerly co-directed by Dr. Bruce Hoffman, Rand Corporation. These centers conduct terrorism research and provide portals as a service for academics, journalists, policymakers, students, and the general public. Terrorism research centers' portals are primarily providing information retrieval and dissemination services except for a few organizations such as the Terrorism Research Center (TRC) and the National Memorial

Institute for the Prevention of Terrorism (MIPT) that have expanded their functions to

include personalization (TRC) and the Emergency Responders Knowledge Base (MIPT).

For example, the TRC, founded in 1996, has the highest number of portal features

(31/61), including four terrorism databases, and is highly recommended with about 5,000 incoming links [128].

In Kennedy and Lum’s empirical study [123] of how criminal justice scholars can expand their research to the terrorism domain, they generated lists of organizations conducting terrorism research and 28 different datasets. Using their list of terrorism organizations and Carnegie Mellon University’s portal taxonomies [129], the terrorism web portals were examined to identify their features, types of information such as databases of terrorists’ Websites, incident databases, and integrated applications that are available

[130].

For thirty of the 97 portals, error messages were received and new URLs could not be identified. For thirteen other portals, terrorism information could not be located. This may be associated with the dynamic nature of the Web and the fact that content is constantly being removed from organization’s web servers. For the remaining 54 terrorism portals, we identified the types of information available and features of a terrorism information portal using CMU’s informational dimensions such as collection, application, and value-added applications, and the percentage of terrorism research organizations providing the services. For the 54 terrorism portals, the information was divided into two categories: unstructured and structured information. For unstructured terrorism information, 74% were documents (full-text); 54% were links to external

resources; 48% were educational resources; 39% were news; and 28% were

Congressional testimonies. Structured information focused on terrorism incident data

(40%). None of the web sites provide access to or maintain information generated by terrorists such as their Websites or forums.

Terrorism research centers' portals provide access to a diversity of unstructured (e.g., reports, news stories, transcripts) and structured (terrorism incident database) information but fall short of analysis tools for integrating the resources and supporting information fusion (including post-retrieval analysis). After searching the terrorism portals, returned results are presented as lists of ranked URLs. The user has to manually browse through the lists to locate relevant resources and establish relationships among the documents.

Since terrorist groups are from all over the world, the language barrier problem has to be addressed to study the multilingual Dark Web data. Terrorism research centers’ portals mainly focus on maintaining and providing access to terrorism incident database. The collections being archived by these portals are often language specific, thus restricting counterterrorism researchers to from obtaining a comprehensive understanding of Dark

Web information in different languages.

4.1.3 Multilingual Issues

Terrorism is an international issue and terrorism-related information is in various

European, Asian, and Middle Eastern languages. However, language barriers prevent effective and efficient discovery of terrorism intelligence. The broad diversity of the

terrorism-related Websites presents a substantial research challenge in the field of information retrieval.

Multilingual Information Retrieval (MLIR), responding to a query by searching for documents in more than one language, has been studied to solve the problem. Most reported approaches translate queries into the document language, and then perform monolingual retrieval. There are three main approaches in query translation: using machine translation, a parallel corpus, or a bilingual dictionary. The machine translation- based (MT-based) approach uses existing machine translation techniques to provide automatic translation of queries. A corpus-based approach analyses large document collections (parallel or comparable corpus) to construct a statistical translation model.

Performance relied largely on the quality of the corpus. Parallel corpus is very difficult to obtain, especially for certain domains such as terrorism. In a dictionary-based approach, queries are translated by looking up terms in a bilingual dictionary and using some or all of the translated terms. This is the most popular approach because of the wide availability of machine-readable dictionaries. Various techniques have been proposed to reduce the ambiguity and errors introduced during query translation. Among these techniques, phrasal translation, co-occurrence analysis, and query expansion are the most popular.

4.2 Research Questions

We will use the ASKE approach to design and develop a knowledge portal to address the challenges in the counterterrorism domain. In this study, we aim to address the following research questions:

1. How can intelligent collection building techniques proposed in ASKE framework be

used to build and maintain a high-quality, up-to-date collection of Dark Web data,

and help resolve the information management problems in counterterrorism research?

2. How can information searching, semantic search, text mining, post-retrieval analysis,

and Cross-Language Information Retrieval techniques help counterterrorism

researchers efficiently access, analyze, and understand the Dark Web collection?

3. How can artificial intelligence and visualization techniques be applied on the Dark

Web collection to enable better understanding of the terrorism phenomena and

support the knowledge creation and discovery patterns in counterterrorism research?

The remainder of the paper presents our work in studying these three questions.

4.3 Implementation of Dark Web Portal

To study the research questions above, we build an intelligent Web portal called Dark

Web Portal based on ASKE framework to assist counterterrorism experts to locate, collect, access, analyze, and manage Dark Web data. Through the Dark Web Portal, experts will be able to quickly locate specific Dark Web information in the data collection through keyword search and semantic search. To cater to the nature of terrorism phenomena and the requirements of counterterrorism experts, a key component

“the integration of multilingual information resources” is added to the ASKE framework.

4.3.1 Dark Web Data Collection Building

100

Since the post September 11 war on terrorism has increased terrorist groups' dependence on the Internet and their "born digital" content is at-risk, it is imperative to assure long- term research access to multilingual terrorist digital information. As we discussed before, neither manual selection method nor automatic Web crawling method is appropriate to build a Dark Web archive with both high precision and high coverage.

Based on the approach proposed in ASKE, there are seven major steps in collecting

Terrorism Websites which were mainly done through systematic and careful semi- automatic work.

• Identify terrorist groups from reliable sources: retrieve terrorist groups’ detail

information, such as terrorist group names, leaders, etc. from reliable sources, for

example, the terrorism reports of the governments of each country;

• Identify terrorist group URLs and forum URLs: manually identify the URLs created

by the terrorist groups, e.g., using terrorist group names and terrorism keyword

lexicon in native language to search the general purpose search engines; adopt these

URLs as the starting URLs, and apply the method discussed in ASKE framework to

extract the favorite links, out-links, and back-links, then after filtering expand the

starting URLs; often the identified terrorist group Web sites contain forum sections or

provide links to their major forums.

• Identify forums hosted on public ISPs: besides forums on extremists’ own Web sites,

many extremist forums are hosted on public ISP servers such as Yahoo! Groups and

Google Groups, AOL Groups, MSN Groups, and other regional ISPs. We identify a

101

list of these popular public ISPs and search for relevant forums using the terrorism

domain lexicon we created in the previous step; by browsing through the messages of

the forums returned from our search, relevant extremists’ forums can be extracted;

• Filter the identified Web sites: The Web sites and forums identified from the previous

steps are filtered by domain experts to make sure that irrelevant or bogus sites do not

make way into our final collection. After the filtering, contents from the identified

Web sites are ready to be collected in the next step.

• Collect Web content generated by terrorist groups: automatic Web crawler is applied

here to collect digital content generated by terrorist groups in all formats, including

Web page format, Text format, and Media format; the Web crawler also collects meta

data from forums, such as authors, headings, postings, threads, time-tags, etc.;

• Generate multilingual Dark Web collection: to make the Web content collected in the

last step searchable for information seekers, each document is indexed into words,

and the relationships between the words and the documents are recorded; various

information retrieval techniques, such as stemming and stop-word removal, can be

applied if necessary; the resulting searchable indexes are then stored into a database

for document retrieval and further analysis;

• Build semantic Dark Web collection: the semantic Dark Web collection is built to

support semantic search and analysis; the terrorism ontology is a way of organizing

data and establishing relationships between concepts; we uses an existing one on the

web at http://www.mindswap.org/dav/ontologies/terrorism.owl , which contains

seventy different classes, and 173 properties (71 DatatypeProperties and 102

102

ObjectProperties); the semantic collection is generated via the tool “Metadata

Extractor”.

To generate a comprehensive picture of Dark Web for counterterrorism researchers, our

Dark Web collection is complemented by collecting the information generated by terrorist groups in different regions (Eastern Asia, United States, South America, and

Europe, etc.) and different languages (Arabic, English, Spanish, etc.). The process from identifying group URLs to generating a multilingual/semantic Dark Web collection will be repeated periodically to keep the collection up-to-date. Also, the information collected during different time periods can be analyzed to study the dynamic evolution of the terrorist groups over time. In Dark Web portal, we collect terrorist group Web sites and

U.S. domestic extremist forums separately. We will report the details of these collections in the following sections.

1. Dark Web collection for Terrorist group Web sites

Our goal is to build a high-quality, up-to-date Dark Web collection which contains multilingual information created by major terrorist groups in the world. To keep the Dark

Web collection up-to-date, we conducted two batches of Dark Web collection building in

April 2004 and June 2004.

In April 2004 we started the process by identifying the groups that are considered by authoritative sources as terrorist groups. The main sources we used to identify US domestic extremist groups include: Anti-Defamation League (ADL, http://www.adl.org/learn/ext_us/default.asp), FBI

103

(http://www.fbi.gov/congress/congress02/jarboe021202.htm and http://www.fbi.gov/congress/congress01/freeh051001.htm), Southern Poverty Law

Center (SPLC, http://www.splcenter.org/intel/history.jsp), Militia Watchdog (MW, http://www.militia-watchdog.org/m1.htm), and Google Web Directory (GD). To identify international terrorist groups we relied on the following sources: United States

Committee For A Free Lebanon (USCFAFL), Counter-Terrorism Committee (CTC) of the UN Security Council (UN), US State Department report (US), Official Journal of the

European Union (EU), and government reports from the United Kingdom (UK),

Australia (AUS), Japan (JPN), and P. R. China (CHN). These sources have been identified following the recommendations of core terrorism authors.

A total of 224 US domestic extremist groups (see Appendix A) and 440 international terrorist groups (see Appendix B) have been identified. Based on the categorization method suggested by SPLC, the US domestic groups were categorized into 7 different categories: Black Separatist (BS), Christian Identity (CI), Militia, Neo-Nazis (NN), Neo-

Confederate (NC), Racist Skinhead (RS), White Supremacy (WS), and others. Figure 4.1 summarizes the number of US domestic groups we identified from each source within each category. From the figure, we can see that the collection covers a large variety of

US domestic extremist groups and could serve as a good resource to study all aspects of terrorism in the US.

104

Figure 4.1 Number of U.S. domestic groups identified from each source within each category

Figure 4.2 Number of international groups identified from each source within each geographical location

105

For international terrorist groups, there is no standard categorization available. The domain expert identified the categories based on literature and her knowledge. In addition, we categorized the international groups based on their primary geographical locations.

Figure 4.2 summarizes the number of international groups we identified from each source within each geographical location. From the figure, we can see that the collection covers terrorist groups from a large span of geographical locations in the world and could serve as a good collection to study cross-jurisdiction terrorism phenomena.

The spider agents are deployed as sensors in the architecture of social sensor networks.

Figure 4.3 depicts the structure of the sensor network for U.S. domestic extremist groups.

The sensor network consists of two different levels of networks: the websites and categories that have higher activity levels are located in the inner circle (Level 1), e.g.,

“White Supremacy” category; the outer circle (Level 2) is composed the related websites and groups which are extended from the inner circle based on link connections and organization connections. With the sensor network, it becomes possible to sense dynamic nature of U.S. domestic extremist activities in a timely manner. The positions of websites and extremist groups in the structure are dynamically changed as spider agents sense various activities in forums, websites, and groups.

106

Figure 4.3 The social sensor network for U.S. Domestic Extremist Groups

107

Category # of pages US Domestic Black Separatist 121 Christian Identity 36,366 Militia 15,997 Neo Confederate 2,478 Neo Nazis 19,384 Racist Skinhead 10 White Supremacy 12,249 Others 7,692 Spanish Left Wing Paramilitary Group 45 Leftist Group 2 Maoist Rebel Group 3,704 Marxist Insurgency/Guerrilla 540 Separatist Group 35,497 Socialist Group 56,671 Others 0 Arabic Jewish 1,498 Secular 1,062 Shi'a Muslim 916 Communist/Socialist 1,291 Secular 6,185 Sunni Muslim 252,177 Others 6,495 Table 4.2 Number of pages collected for the terrorist groups within each category For the first batch, we manually identified the URLs of the terrorist groups’ web sites from the reports alluded to by the sources mentioned above then searched the web using the group names in their native language as queries. To ensure that our collection covers all the major regions in the world, we sought the assistance of language experts in

English, Arabic, Spanish, Japanese, and Chinese to help us collect URLs in different regions. At this time, we identified 114 URLs from US domestic extremist groups, 48

URLs from Spanish-speaking terrorist group, and 66 URLs from Arabic-speaking groups.

After the URL of a group is identified, we used the SpidersRUs toolkit

(http://ai.bpa.arizona.edu/spidersrus), a multilingual Digital Library building tool, to

108

collect all the static text-based Web pages (html, txt, pdf, and doc) under that URL. So far, we have collected 500,000 Web pages created by US domestic groups, 300,000 Web pages created by Arabic-speaking groups, and 100,000 Web pages created by Spanish- speaking groups. Table 4.2 demonstrates the number of Web pages for the terrorist groups within each category.

US Domestic Collection # of Files Volume (Bytes) Total 396,105 27,630,493,405 Indexable Files 316,692 9,288,983,289 HTML Files 102,168 3,761,303,743 Word Files 425 301,294,350 PDF Files 1,201 947,561,966 Dynamic Files 207,985 2,959,621,213 Text Files 4,112 793,433,495 Excel Files 2 140,800 PowerPoint Files 7 1,035,738 XML Files 792 524,591,984 Multimedia Files 70,832 15,478,613,165 Image Files 65,296 1,214,302,446 Audio Files 4,898 11,647,742,758 Video Files 638 2,616,567,961 Archive Files 767 409,018,557 Non-Standard Files 7,814 2,453,878,394 Table 4.3 Documents spidered in the second batch for US Domestic Extremist Groups In June 2004 the second batch collection was built by expanding and updating the first batch collection. More authoritative sources are adopted to identify terrorist groups, such as a report from Dartmouth College (https://www.ists.dartmouth.edu/TAG/cyber- capabilities-terrorist.htm). Through automatically extracting out-going URLs from favorite links of the terrorist groups’ URLs that we have identified in the first batch and manually filtering, 386 US domestic, 151 Arabic, and 83 Spanish terrorist URLs were included in the second batch collection. All types of documents were collected in the second batch collection. As well as the static text-based files (html, txt, pdf, and doc)

109

included in the initial collection, the second batch collection also includes: Dynamic text- based files (asp, php, cgi, etc), Media files (images, video and audio files), Archive files

(zip packages, rar packages, etc), and Non-standard file types. Tables 4.3-4.5 present the detail statistics of the Web documents spidered for US domestic extremist groups,

Arabic-speaking terrorist groups, and Spanish-speaking terrorist groups respectively.

Table 4.6 summarizes the comparison between the 1st and 2nd batch collection.

Arabic Collection # of Files Volume (Bytes) Total 222,687 12,362,050,865 Indexable Files 179,223 4,854,971,043 HTML Files 44,334 1,137,725,685 Word Files 278 16,371,586 PDF Files 3,145 542,061,545 Dynamic Files 130,972 3,106,537,495 Text Files 390 45,982,886 PowerPoint Files 6 6,087,168 XML Files 98 204,678 Multimedia Files 35,164 5,915,442,276 Image Files 31,691 525,986,847 Audio Files 1,973 3,750,390,404 Video Files 733 1,230,046,468 Archive Files 1,281 483,138,149 Non-Standard Files 7,019 1,108,499,397 Table 4.4 Documents spidered in the second batch for Arabic-Speaking Terrorist Groups

Spanish Collection # of Files Volume (Bytes) Total 332,134 6,207,859,955 Indexable Files 281,382 4,421,338,803 HTML Files 154,671 2,505,598,128 Word Files 389 21,549,415 PDF Files 156 34,793,876 Dynamic Files 125,586 1,846,283,672 Text Files 556 9,829,425 Excel Files 2 71,168 PowerPoint Files 7 29,34,272 XML Files 15 2,78,847 Multimedia Files 44,671 1,6098,93,751 Image Files 44,284 9696,82,787 Audio Files 316 5715,19,457

110

Video Files 71 686,91,507 Archive Files 110 229,34,999 Non-Standard Files 5,971 1536,92,402 Table 4.5 Documents spidered in the second batch for Spanish-Speaking Terrorist Groups

US Domestic Spanish Arabic Initial 2nd Initial 2nd Initial 2nd Batch Batch Batch Seed Total 81 386 37 83 69 128 URLs From 63 266 0 0 23 31 Literature & Reports From Meta- 0 0 37 48 46 66 search From out- 18 120 0 32 0 31 link extraction Terrorist Groups 94 219 7 10 34 36 Collection Total 125,610 396,105 106,459 222,687 322,524 332,134 Size Multimedia 0 70,832 0 35,164 0 44,671 Files Table 4.6 Comparison between the 1st and 2nd batch collection

2. Dark Web collection for US domestic extremist forums

Following the ASKE framework, we started our forum collection from identified 386 U.S. domestic extremist groups URLs. Using the information of these groups as queries, we searched major search engines and public ISP Web sites (e.g., Yahoo! Groups, Google

Groups, etc.) for forums created and maintained by the identified extremist groups. After the expansion and filter steps, we identified a total of 105 extremist forums of which 12 are hosted on stand-alone extremist Web sites, 47 are hosted on Google Groups, 31 are hosted Yahoo! Groups, ten are hosted on MSN Groups, and five are hosted on AOL groups (see Appendix C). Figure 4.4 provides a summary and categorization (based on

SPLC) of the forums we identified.

111

As shown in Figure 4.4, the identified forums were created by a wide variety of U.S. extremist groups and should serve a comprehensive resource for domestic extremist movement studies.

35 30 25 Neo -Nazis 20 White Supremacist 15 Black Separatist 10 5 Christian Identity 0 Militia Neo-Confederate Others

Figure 4.4 Summary and categorization of identified U.S. domestic extremist forums After obtaining membership for the password-protected forums, we spidered data from the identified extremist forums. Table 4.7 is a summary of the number and volume of different types of documents we downloaded from the extremist forums.

As we can see from Table 4.7, textual files (e.g., HTML files, PDF files, Word Files,

Excel Files, etc.) are the largest category on extremist forums. Multimedia files also comprise a significant presence on extremist forums. We found that the usage pattern of multimedia on standalone extremist forums is quite different from that on extremist forums hosted on pubic ISP servers. The average number of multimedia files posted on standalone extremist forums (1195.4) is much larger than that of the public ISP hosted

112

extremist forums (70.8). However, even though a smaller number of image files are posted on the public ISP hosted forums, they are much larger in volume. For audio and video files, the ones that were posted on standalone forums are larger both in terms of quantity and volume. The different levels of multimedia usage in the two types of forums could be resulted from the limitations on multimedia use that the public ISPs usually set in their forums.

Stand Alone Forums Public ISP Forums

# of Volume # of Files Volume Files (Bytes) (Bytes)

Total 116,419 7.7G 524,652 20G

Textual 93,655 6.5G 350,046 10.7G Files

Multimedia 21,518 1.1G 6,511 1.3G Files

Image 21,177 374M 5,393 1.06G Files

Audio 107 405M 589 186M Files

Video 234 358M 529 21M Files

Non- Standard 1,246 45M 168,095 9G Files

Table 4.7 Summary of Document Types in the Forum Collection Another difference between the two types of forums is the use of non-standard files. We found the amount of non-standard files on the public ISP forums was much larger than the stand alone forums. These non-standard files could be files that can only be opened

113

by using some obsolete software or they can be encrypted materials that were deliberately made inaccessible for normal software.

As described in ASKE framework, our spider agents automatically extract important information such as number of participants and postings from the forum files. Figure 4.5 gives an example of information that has been extracted, such as userid, user rank, posting time, thread title, posting message, reference message, etc. This information can be used in various trend analysis and temporal analysis in the future.

We found that the domestic extremist forums are very popular. In our collection, the average number of participants of each forum is 788.5 and the average number of posts on each forum is 6461.6. The largest forum in the collection is the Neo Nazi forum

StormFront.org with 55,834 registered participants, 189,816 existing threads, and

1,923,169 postings. The most popular category is Neo Nazi forums with average numbers of 2234.4 participants and 21,217 posts per forum. The second most popular category is the White Supremacy forums with numbers of 80.31 participants and 336 postings per forum. The smallest category is the Eco Terrorism forums with, on average, 15 participants and 147.6 postings per forum.

Through statistical analysis, we found that the numbers of participants and postings on the extremist forums follow a power-law distribution:

−γ p(n) ~ n where n is the number of participants or postings on a forum; and p(n) is the probability of a forum having n participants or postings.

114

Figure 4.5 An example of U.S. domestic extremist forums Regression analysis showed (see Figure 4.6) that the number of participants on extremist forums follows a power-law distribution (R = 0.92) with exponent g = 0.339 and the number of postings on extremist forums follows a power-distribution (R = 0.78) with exponent g = 0.210. Such distributions indicate that “preferential attachment effect” may have affected the development of extremist forums. The more popular forums tend to

115

attract more participants and become more and more popular; while the less popular forums are likely to remain unpopular. Thus, monitoring the most popular extremist forums is very important since those forums are where most future participants go and where ideas are expressed and contacts made.

Figure 4.6 The distributions of number of participants and number of postings on extremist Forums

4.3.2 Post-retrieval Analysis and Multilingual Support

To address the information overload, the Dark Web Portal is fitted with post-retrieval components, including categorizer, summarizer, and visualizer.

Document Summarizer. Automatic summarization has been applied as a document preview tool in many information retrieval systems [117]. In the Dark Web Portal, a multilingual Summarizer was developed based on the AI Lab TXTRACTOR [102] that uses sentence-selection heuristics to rank text segments. This heuristic strives to reduce redundancy of information in a query-based summary [131]. It supports summarizing

116

Web documents in Arabic, English, and Spanish at the same time. The summarization involves three components: 1) sentence evaluation, 2) segmentation or topic boundary identification, and 3) segment ranking and extraction. First, a sentence evaluation component parses the original Web page and extracts all sentences. These sentences are evaluated based on linguistic heuristics including presence of cue phrases (e.g. "in summary", "therefore" in respective languages), tf*idf score normalized for the sentence length, sentence position, and sentence length. Second, the Text-Tiling algorithm [132] is used to analyze the Web page and determine where the topic boundaries are located. The

Web page is thus segmented into its main topics. A Jaccard similarity function is used to compare the similarity of different blocks of sentences. Third, the Summarizer ranks the document segments based on the scores given to the sentences and extracts high-ranking sentences from different segments as summary sentences.

Document Categorizer. The document categorizer organizes the returned Web documents into 20 or fewer different folders labeled by the key phrases appearing most frequently in the page summaries or titles. Key phrases with high occurrences in the returned results are extracted as folder topics. Web documents that contain a folder topic are included in that folder. One Web document may appear in several folders if it contains multiple folder topics.

For Web documents in Arabic, when the categorizer is invoked all the returned results are processed and key phrases that appear in the titles and summaries are extracted by matching to a phrase lexicon in the respective language. To create the lexicons, we

117

extracted meaningful phrases from the Dark Web collection by using the Mutual

Information program, which is based on a statistical method [114]. The method is an iterative process of identifying significant lexical patterns by examining the frequencies of word co-occurrences in a large amount of text.

For Web documents in English or Spanish, we use the Arizona Noun Phraser (AZNP) [93] to extract meaningful phrases from the titles and summaries of the search results.

Developed at the Artificial Intelligence Lab of the University of Arizona, AZNP extracts all the noun phrases from each Web page automatically based on part-of-speech tagging and linguistic rules. An indexing program calculates the frequency of occurrence of these phrases and selects the 20 most frequently occurring phrases to index the results.

In our document categorizer we are using only titles and summaries to extract keywords since it is practical and permits dynamic categorization. Previous research has shown that clustering based on snippets is almost as effective as clustering based on a whole document [113].

Document Visualizer. The Dark Web Portal also supports visualizing the retrieved Web documents and helps reduce information overload when a large number of search results are returned. Using the document visualizer, counterterrorism researchers can obtain a meaningful and comprehensive picture of a large number of search results. The document visualizer provides two types of visualizations: the Jigsaw and Geographic Information

Systems (GIS).

118

The Jigsaw is a two-dimensional map generated by using the Kohonen self-organizing map (SOM) algorithm. SOM is a two-layered neural network that automatically learns from the input Web documents and clusters them into different naturally occurring groups, and it has been used in image processing and pattern recognition applications. In the Jigsaw map, similar documents are assigned to adjacent regions which are labeled by key phrases identified by the AZNP or the mutual information program. The size of a region on the map indicates how many pages are assigned to it.

In the GIS SOM visualizer, Web documents are shown as points on a two-dimensional map with their positions determined by the SOM algorithm. The map's background shows contour lines representing the varying values selected by users (e.g., frequency of occurrence of query terms in the Web pages) and is independent of the points' positions.

Users can navigate on the map by clicking on the buttons and resize a certain part of the map by dragging a rectangle that will highlight the set of Web pages listed on the bottom right side of the pop-up window.

Multilingual Support. To support multilingual retrieval of terrorism-related Websites, we proposed to use MLIR and Machine Translation in our Dark Web Portal. A complete

Web-based multilingual retrieval system consists of five components: (1) Spiders (a.k.a.

Crawlers) to retrieve Web pages by recursively following URL links, (2) an Indexer to tokenize multilingual Web pages into words or phrases, (3) a Query Translation Engine to translate the query into document languages, (4) a Retrieval Engine to get relevant

119

results, and (5) a Document Translation Engine to translate retrieved documents into users’ familiar languages.

Spiders are used to build multilingual collections. Indexers are designed to work with different language encodings. For languages with a rich morphology such as Arabic, language normalization is frequently used in indexing to improve retrieval performance.

The Translation component is the core of the system. It is responsible for translating search queries in the source language into the target language. We used a dictionary- based approach combined with phrasal translation and co-occurrence analysis for translation disambiguation. In the dictionary lookup process, the entry with the smallest number of translations will be preferred over other candidates. In addition we conducted maximum phrase matching. Translations containing more continuous key words will be ranked higher than those containing discontinuous key words. Co-occurrence analysis also was used to help choose the best translation among candidates. The Retrieval Engine would be similar to those monolingual system retrieval engines since queries are translated into document languages in the previous step. To make such a system useful to end users, a document translation engine is necessary. Commercial machine translation software such as SYSTRAN would serve our needs. With these five components, users are able to retrieve documents in languages other than the query language and understand the content of these documents.

4.3.3 Searching and Browsing in the Dark Web Portal

120

The Dark Web Portal supports searching and browsing terrorist groups in different regions and languages, for example, English, Arabic and Spanish. In particular, the portal provides three versions of user interfaces to accommodate users with terrorist groups’ native languages so that users can get better understanding of the terrorist groups. They look similar and present the same functionalities, except that they use different encoding schemes and languages.

By configuring KFC, users can easily get reports about the interested terrorism information regularly. The following is a KFC example that supports to retrieve dynamic information of an US extremist group “Women for Aryan Unity”.

Dr. John Casey [email protected] KKK, white supremacy http://wau14.com/ http://www.stormfront.org/forum/ http://www.w-a-u- argentina.blogspot.com/ http://www.stormfront.org/crusader/texts/wau/ White Supremacy white people are superior to people of other racial backgrounds white supermacy superior anti-black Racism Ku Klux Klan Women for Aryan Unity

121

reveal new ways of thinking and new manners of acting that will rediscover the mysteries of the race Women for Aryan Unity abortion white children compete with men

0 8 * * * fund activity or parade or magazine schedule

In this KCF, the user “John Casey” would like to make sure 4 URLs are collected and indexed, and scan these sites in different frequency. The URL http://www.stormfront.org/forum/ is an online forum, which will be scanned every 10 minutes. The main website http://wau14.com for the group “Women for Aryan Unity” is checked every day. In the topic section, a category “White Supremacy” is defined, and also the group is defined as a subtopic of the category. In the query section, John tries to get the information of the group about fund raising, different activities, and schedules of these activities. The report with search results related to the topics and defined by this query will be delivered 8am every day.

In the Dark Web Portal, there are two types of search forms available: simple search and advanced search. The default one is simple search. Users may switch between the two using the “Advanced Search” link. For the simple mode (see Figures 4.7.a, 4.8.a), all

122

users need to do is enter one or multiple keywords into the search box, then the portal will find the Web documents which include all of the keywords. For the advanced mode

(see Figures 4.7.b, 4.8.b), the advanced search screen will give users a much greater range of options to choose from to refine the search. In addition to searching Web documents with all of the keywords, users are able to find corresponding Web pages with the exact phrase, or with at least one of the keywords, or without the keywords. There are three additional ways of searching terrorist groups’ information. Users can either restrict the results t within some terrorist categories as we discussed above, or choose different file types (pdf files, Word files). Furthermore, a user can select the time period he/she is interested in. Users may refine the search query with the individual or combined options.

After searching, the top 100 results are returned. On the first result page (see Figure 4.7.c,

4.8.c), the top 20 results are displayed in sorted order of relevancy to the query as measured by tf*idf score. Users can check other results by using the “Next” link at the bottom of the page. There are also search boxes available at the top of each result page to allow users to carry out a quick search. For each query, the first result page displays

“Suggested Keywords”, a set of relevant keywords such that the user can expand or refine the original search query, which are obtained via a Concept Space approach [133-

135]. Additionally users can switch among Webpage, pdf, and Word results by the corresponding format links.

For each result, since these terrorist group Web pages often disappear randomly, cached pages for different time periods are provided. We have built two batches of our Dark

123

Web collection. The system also displays the terrorist group name, the corresponding category, and the country where the terrorist group is located for each Web page. By clicking on the “Sort by Group”, “Sort by Category”, and “Sort by Country” links, users can sort the results by groups, categories, and countries very conveniently and easily locate the terrorism information they need. The summarizer function is also implemented for each result. Users can summarize each result flexibly using three or five sentences by selecting the number of sentences under the result. The summarization result is shown in

Figures 4.7.d and 4.8.d.

By clicking on the “Organize” tab, users can go to the categorizer page where all the results are categorized into folders with extracted topics. Clicking on the folders of interest gives a list of URL titles that appear under the relevant folder for browsing (see

Figures 4.7.e and 4.8.e). To visualize the returned results, two visualizers are provided.

When the “Map” tab is clicked, a new window containing an SOM map is activated

(shown in Figures 4.7.f and 4.8.f). Users can click on a region to see a list of the pages on the right and can open the pages by clicking on the titles. The GIS map can be activated by clicking on “Try our new visualization tool” (see Figures 4.7.g). Users can navigate on the map by clicking on the buttons and resize a certain part of the map by dragging a rectangle that will highlight the set of Web pages listed on the bottom right side of the pop-up window.

In addition, the portal provides Web directories for US domestic groups, Arabic-speaking terrorist groups, and Spanish-speaking terrorist groups (Figures 4.7.h and 4.8.h). These

124

Web directories are organized in a hierarchical structure and compiled based on the categorization method suggested by SPLC and our domain expert’s knowledge. By traversing the predefined Web directory hierarchies, users may easily locate the terrorist groups or their supporters’ websites.

4.3.4 Multilingual Support

To support multilingual retrieval of terrorism-related Websites, the Dark Web portal helps English-speaking experts access, analyze, and understand Dark Web information in

Arabic languages.

Arabic is a morphologically rich language. In order to handle the variations in the way text can be represented in Arabic, we performed normalization on Arabic Web pages including stemming and stopword removal. In stemming, Arabic terms were transliterated into Roman characters. Prefixes and suffixes on the transliterated terms

, ال , ة , ,م ,ل ,ي ا : were removed using a heuristics-based approach. Removed prefixes are

The stemmed . وا , , و , ن , و , , , , ,آ ,ه ه , ,ه , , , :and suffixes are transliteration was converted back to Arabic. Since most Arabic Web pages often omit diacritics (weak vowels) and only preserve letters of alphabets, diacritics were removed to ensure consistency. A stopword list of 347 words was originally taken from the

University of Neufchatel’s (http://www.unine.ch/info/clef/) repository. One hundred more words were manually added from the training process of Mutual Information and our final stopword list contains 550 words.

125

We used the dictionary-based approach in query translation. Two dictionaries were used at this stage: an English-Arabic dictionary constructed from Al-Misbar, an online dictionary (more than 20,000 entries), and an Arabic-English dictionary from Tuft’s

University (50,000 entries). We obtained more than 60,000 entries when combining two dictionaries and diacritics were removed in dictionary entries. The collection was first indexed against the combined dictionary. Co-occurrence scores between every two dictionary terms were calculated and stored in the database. In order to do query expansion, we integrated the Arabic Stemmer with the Mutual Information program to extract meaningful Arabic phrases. After training, we extracted 20,383 entries (Arabic phrases) from the collection in Arabic. These phrases were used as potential query expansion phrases. Pseudo relevance feedback was used to perform query expansion. Our retrieval model is a standard tf*idf based model.

Figure 4.9 shows a sample user session of the Dark Web Portal. After the user provides an English query (Figure 4.9.a) the system will first apply a word-by-word translation which shows all the possible translations of a word (Figure 4.9.b). In the phrasal translation phase, the system detects multiple English words as a phrase, based on a phrase dictionary, and translates them into one Arabic word/phrase (Figure 4.9.c). In Co- occurrence Analysis, the system determines the best combination of possible translations based on their co-occurrence in a domain-specific Arabic document corpus. The best translation combination is highlighted among all the possible translations (Figure 4.9.d).

The best translation combination is used to retrieve Arabic documents and the results are displayed in Figure 4.9.e. In the last step, retrieved Arabic documents are translated back

126

to English, the original query language, using the machine translation system (Figure

4.9.f).

4.3.5 Semantic Search in the Dark Web

Although the semantic search engine is still in its infancy in Dark Web Portal as the ranking algorithm has not yet been fully investigated, it produces encouraging results.

Figure 4.10 shows the results when we search the famous terrorist group “Roubaix Gang” using semantic search engine. Behind the scene, the collections are annotated by the metadata extractor and thus being associated with semantic mark-ups.

As shown in the figure, the document which mentions the group name “Roubaix Gang” and its related activities appears in the top, as it gets the biggest relation weight. The documents that mention other semantic entities (e.g. members of “Roubaix Gang”) with which many terrorist members have relations also get good rankings, since in the ASKE framework we give implicit relations half the weight of the explicit ones. With this profound semantic search mechanism, we can make use of the available semantic relations between different resources to bring forward most relevant information.

127

a. Simple Search Page b. Advanced Search Page

d. Web page Summarization Results c. First Result Page

f. Visualizer - SOM e. Web Page Categorization Results

h. Web Directory g. Visualizer - GIS

Figure 4.7 The screenshots of Dark Web Portal for U.S. domestic groups

128

a. Simple Search Page b. Advanced Search Page

c. First Result Page d. Web page Summarization Results

e. Web Page Categorization Results

f. Visualizer - SOM

h. Web Directory

Figure 4.8 The screenshots of Dark Web Portal for Arabic-speaking terrorist groups

129

a. Search Page: User types in English Query

b. Word-by-word Translation: System lists all possible translations

c. Phrasal Translation: The system detects and translates multi-word concepts in the original query as phrases

d. Co-occurrence Analysis: The system determines the best combination of possible translations based on their co-occurrence in an domain-specific Arabic document corpus.

e. The system retrieves: Arabic documents using the translated query and provides English translations of the titles and summaries of the results.

f. The system provides English translations of the titles and summaries of the results using a machine translation system.

Figure 4.9 Jihad multilingual portal user sessions

130

Explicitly Ranking mention

Members of Roubaix Gang

Figure 4.10 The screenshot of semantic search for the keywords "Roubaix Gang"

4.4 Conclusions and Future Directions

Information overload, uncertain data quality, and lack of access to integrated datasets and advanced methodologies for studying terrorism are major hurdles and challenges which both traditional and new counterterrorism researchers have to overcome. In this chapter we reported the current status of digital archiving for terrorists’ resources, reviewed

131

terrorism research portals by specialized research centers, and analyzed contemporary technology solutions to help fill in existing gaps. By using the ASKE framework, we have proposed an approach to build an intelligent Web portal, called Dark Web Portal, to assist counterterrorism experts to locate, collect, access, analyze, and manage Dark Web data.

We have discussed the development of Dark Web Portal, which is aimed at supporting the counterterrorism research community. Two batches of collections which cover US domestic groups, Arabic-speaking terrorist groups, and Spanish-speaking terrorist groups have been implemented. We believe that it will facilitate better understanding of the global terrorism phenomenon and provide a systematically integrated research tool for culturally diverse counterterrorism researchers.

In this chapter, we confirm that ASKE framework is an applicable and efficient approach to build domain/application specific web portals. The semantic search facility developed within the framework takes advantage of semantic representation of information to facilitate keyword searching. It produces encouraging results. More further investigation need to be conducted: i) a more fine-grained ranking algorithm which gives appropriate consideration for all the factors a ﬀecting the search results, and ii) the e ﬀect of implicit relations on search results.

During the process of building Dark Web portal, we are aware of a number of limitations associated with this framework. For example, the manual specification of mappings in the process of setting up the metadata extractor makes the approach heavy to launch.

132

CHAPTER 5

MONITOR FILE SHARING IN P2P WORLD

In 2006, the Motion Picture Association of America claimed that the industry was losing

$6.1 billion annually in global wholesale revenue because of the impact of pirated DVD movies and Internet downloads, based on the study conducted by LEK Consulting LLC

[136]. Although we have no details about how the estimates are calculated, it is apparent that movie piracy has kept increased recently and people are sometimes happy to get movies for free even by violating intellectual property rights. The similar situation happens in music industry, publication industry, and software industry also.

Among six types of common piracy ways (Optical disc piracy, Videocassette piracy,

Internet piracy, Signal theft, Broadcast piracy, and Public performance), the Internet piracy becomes more and more popular as the growth of Peer-to-Peer (P2P) file sharing has been surprisingly fast. It is estimated that currently up to 90 percent of local and 60 percent of backbone traffic is P2P traffic [137]. The financial losses to copyright owners due to P2P can be more severe and cannot be reduced in a short period, as with P2P techniques, it is much easier to distribute copyrighted materials over the Internet in digital forms. Currently there is no efficient way to prevent distribution over P2P networks for copyright holders. Sometimes, pirated materials are distributed over P2P networks even before they are official released. For instance, the Telesync (TS) version of movies is easily produced in the theatres with a cheap camcorder. To develop an effective anti-

133

piracy strategy to fight piracy over P2P networks, we have to understand the details of how pirated materials are generated and disseminated in P2P world.

In this chapter, we will describe how to build an anti-piracy system based on the ASKE framework, generating insights of P2P download activities. The remainder of this chapter is structured as follows: Section 5.1 presents the literature review; Section 5.2 proposes a feasible approach to implement the components in ASKE for different P2P protocols, including the challenges and problems during the process we implement them; we report the services provided to copyright owners and present a case study in Section 5.3 and the conclusions are given in Section 5.4.

5.1 Literature Review

In this section, we review P2P history, popular P2P networks, and the related research in

P2P domain.

5.1.1 P2P History

Since Napster started its service in 1999, P2P file sharing has grown to the point of becoming a significant and growing component of Internet traffic [54, 59].

P2P networks are the virtual networks of computers, in which there are only peers, instead of servers and clients in the traditional networks. Every computer or machine is a peer with the same functionality, although they may have big differences with respect to hardware conditions, operating systems, or connection speed, etc. In traditional networks,

134

each node has its own operations, and the servers are definitely authorities. In contrast, in

P2P networks, there is no centralized authority node, and every peer shares the same configuration and connects to each other equally at the application level. As the exception, some P2P networks have some special peers (super nodes) that are used to handle keyword queries.

Although P2P networks became popular only for recent years, the concept of P2P networking emerged in the early period of network systems. Surprisingly, we can treat both ARPANET in the late 60’s and Usenet in the late 80’s as the predecessors of today’s

P2P networks, since they have the similar features of P2P networks, such as distributed, decentralized networks and used for sharing files among equal peers. The use and development of P2P networks is neglected as World Wide Web growed dramatically in the early 90’s. However, a series of new technologies lead to the explosion of P2P applications. First, the popular mp3 (the MPEG-1 Audio Layer-3) encoding [138] which implemented huge data compression, with free mp3 players (e.g., winamp [139]).

Encodings that made considerable reduction for video data possible were also developed after that (e.g., DivX [140], Xvid [141]). Second, high-speed Internet access becomes popular to end users. Third, the Napster network [142] was started in 1999 totally, totally changing the way of file sharing.

Since Napster’s service was forbidden to stop, many different types of P2P networks have been developed. These new P2P networks support millions of users and billions of file transfers. P2P applications have grown to be a considerable and dominant fraction of

135

Internet traffic. In 2000, three popular P2P applications Gnutella, eDonkey2000, and

Freenet were released. As the first decentralized file sharing network, Gnutella considers all connecting peers completely equal. Any peer fails would not affect the network. eDonkey2000 provides client and server software, and server software is used to facilitate keyword search functionality. Freenet was the first anonymity network. In 2001, Kazaa and Poisoned for the Mac was released. Its FastTrack network was distributed, though unlike Gnutella, it assigned more traffic to 'supernodes' to increase routing efficiency. In

July 2001, the LimeWire client and BitTorrent protocol were released. Until its decline in

2004, Kazaa was the most popular file sharing program. From 2002 through 2003, a number of popular BitTorrent services were established, including Suprnova.org, isoHunt,

TorrentSpy, and The Pirate Bay. With the shutdown of eDonkey in 2005, eMule became the dominant client of the eDonkey network. Currently the most popular networks are

BitTorrent via uTorrent, Gnutella via Limewire, and the eDonkey network via eMule.

Most P2P file sharing applications are implemented satisfying the following requirements:

• Adaptivity: P2P applications run in dynamic P2P networks, where peers may join

or leave the network unpredictably and frequently. Users don’t need to know the

details of the network, even the resources are constantly changed.

• Performance and Scalability: the performance should stay stable when the number

of nodes are increased dramatically, such as constant response time, linear growth

of aggregate storage space, and constantly increasing search.

136

• Reliability: it should have the ability to defense external attacks without

significant data or performance loss.

• Anonymity: it keeps every peer’s private information secured.

5.1.2 P2P Networks

This section reviews the three most popular p2p networks: eDonkey2000, Gnutella, and

BitTorrent.

1. eDonkey2000

The eDonkey2000 (nicknamed “ed2k”) file sharing protocol is implemented by the original windows based eDonkey2000 client [143] developed by MetaMachine, using the

Multisource File Transfer Protocol, and additionally by some open-source clients like

MLdonkey [144] and eMule [57].

The ed2k network is a de-central hybrid peer-to-peer file sharing network with client applications running on the end-system that are connected to a distributed network of dedicated servers. The ed2k protocol uses two TCP ports (4661, 4662) and one UDP port

(4665) default. Data are transferred via TCP, and control packets (such as search related packets) are transferred by either TCP or UDP.

Contrary to the original Gnutella protocol it is not completely de-central as it uses servers; contrary to the original Napster protocol it does not use a single server (farm) which is a single point of failure, instead it uses servers that are run by power users and offers mechanisms for inter-server communication. Unlike super-peer protocols like KaZaa, or

137

the modern Gnutella protocol, the ed2k network has a dedicated client/server based structure. The servers are slightly similar to the KaZaa super-nodes, but they are a separate application and do not share any files, only manage the information distribution and work as several central dictionaries which hold the information about the shared files and their respective client locations.

In the ed2k network only clients share data, as severs index their shared files when servers are connected. When a client starts to download a file or a part of a file, firstly it connects to a server via TCP or sends a short search request via UDP to one or more servers to get the necessary information about other clients offering that file.

The ed2k network is using 16 byte MD4 hashes to (with very high probability) uniquely identify a file independent of its filename. The implication for searching is that two steps are necessary before a file can be downloaded in the ed2k network. First, a full text search is made at a server for the filename, it is answered with those file hashes that have a filename associated which matches the full text search. In a second step, the client requests the sources from the server for a certain file-hash. Finally, the client connects to some of these sources to download the file.

A typical ed2k link includes the filename and the filesize besides hash codes. An example

(a link to the 645.1Mb TV show “The pacific” ep1) is provided below: ed2k://|file|The.Pacific.Part.1.Chi_Eng.HR-

HDTV.AC3.1024X576.x264.mkv|676471938|9a617a5c62a9e2294f0be9b10704c895|h=zunwifgbixqr5kgrtj xkoctaf2wxegub|/

138

The "file" part indicates that this is a file link, as opposed to a server link. An optional

AICH top hash |h=H52BRVWPBBTAED5NXQDH2RJDDAKRUWST| is also included to help recover the file in case of corruption during file transfer.

In the eMule client, ed2k data packets are transferred as blocks of 10,240 bytes in 8 successive packets. All ed2k control packets are wrapped in a special header that starts with the eight-bit character 0xe3, followed by an unsigned long that specifies packet length. On the other hand, ed2k control packets start with a different eight-bit character,

0xc5.

2. Gnutella

The Gnutella protocol [56] is an open, decentralized search protocol for file sharing.

Specifically, it uses HTTP packets for file transfers. Gnutella uses port 6346 and 6347 for

TCP and UDP traffic as default reports, and users can change the ports in Gnutella clients easily. Known Gnutella clients include Limewire, phex, BearShare, and Morpheus. The block sizes depend on each Gnutella client, e.g., Morpheus uses 32,768 bytes, and

Limewire uses 100,010 bytes.

Gnutella nodes work as SERVers and cliENTS at the same time, thus are called servents.

Via client software, users can send out keyword queries and retrieve search results, and in the other way, accept search requests from other servents, check the search keywords against their local files, and respond with appropriate and related results.

Gnutella uses bootstrap way to connect a node to the network. A new servent initially connects to several known hosts that are always connectable to join the system. Once this

139

new servent has one or more open connections with servent already in the network, the servent sends messages to interact with each other to connect more servents in the network. Messages are broadcasted (sent to all servents) or simply back-propagated (sent on the reverse of the path taken by an initial message). To send broadcasted/back- propagated messages, each message has a randomly generated and unique identifier, TTL

(time-to-live) field, and “hops passed’ field, and in the mean time, each servent keeps a short memory of the recently received messages, used to prevent re-broadcasting and implement back-propagation.

There are mainly three types of messages in the network. Group membership (PING and

PONG) messages are broadcasted to announce a new node joining the network, and also share IP addresses and the number and size of shared files. Search (QUERY and QUERY

RESPONSE) messages are broadcasted to the network including user specified search keywords. Each receiving servents matches the keywords against to the local shared file list. When matches are found, QUERY RESPONSES are back-propagated to the requestor including necessary information to download a file. File Transfer (GET and

PUSH) messages are encapsulated as HTTP packets, which are used to file downloading.

As we mentioned above, Gnutella network is dynamically changed, where nodes join and leave the network very frequently. To understand the current Gnutella network, a node periodically PINGs its neighbors to discover other active nodes. By this mechanism, a node which drops connections because of unstable network can always reconnect back to the Gnutella network.

140

3. BitTorrent

BitTorrent [53] is a P2P network targeting to leverage the upload bandwidth of the download peers and achieve fast and efficient large files sharing. During a file transfer in a BitTorrent network, the file is divided into equal-sized blocks (typically 32-256 KB in size), and nodes can download different blocks from multiple peers simultaneously.

BitTorrent uses 6881 as the default port, but most BitTorrent clients support transfer files on arbitrary ports, such as µTorrent, rtorent, Azureus, and Transmission etc.

Unlike Gnutella networks, BitTorrent networks have some special set of nodes, called

Trackers, which keep track of the nodes currently in the system. The tracker receives updates from nodes when nodes join or leave the torrent or periodically (i.e., every 30 minutes), and also sends nodes which share the same file to the node requests the chunks of the file. The tracker information is included in a “.torrent” file, which is used to identify a file to be shared on BitTorrent networks.

Nodes in the BitTorrent network can be divided into two groups based on the percentages of the file the node holds. If nodes have a complete copy of the file and share the file to others, these nodes are called seeds. Otherwise, nodes that only hold part of the file and are still downloading the file are called leechers.

The basic process for a new node to download a file is as the following: a) when a new node loads the torrent file, it tries to contact the tracker to retrieve a list containing a random subset of the nodes that share the file completely or partially; b) the new node then attempts to establish connections to about the nodes in the list got from the tracker;

141

the node can contact the tracker periodically to obtain additional peers; c) the new node now tries to download blocks from and upload blocks to the peers it connects to; the blocks which are least replicated among its neighbors will be downloaded first.

5.1.3 Related P2P Research

To fight digital piracy, a lot of related P2P research studies have been conducted. They are focused on the following three main categories: a) Impacts on industries and strategies, b) characterization of P2P networks (topology, traffic etc.), c) anonymity and privacy. Most studies have been concentrated to only a limited number of P2P protocols, which generally include eDonkey2000, BitTorrent, and Gnutella because they are easy to access and popular.

The first category consists of a bunch of studies about how P2P technology impact on music, movie, software, publication, and computer games industries, and some other general economic, social, and ethical implications of this technology. Lu [145] discussed several critical issues caused by the current illegal use of P2P technology for sharing copyrighted music: e.g., the serious damage to music production and the infringement on copyright holders’ interests, but in the other hand, he agreed with the significance of P2P as an advanced technology for popularizing music and sharing human and spiritual values with more people. Walls [146] examined the rate of motion-picture piracy across a sample of 26 diverse countries and conducted a cross-country regression analysis which indicated that piracy is increasing in the level of social coordination and the cost of

142

enforcing property rights, unrelated to income and decreasing in internet usage. Andersen and Frenz [147] used representative survey data from the Canadian population collected by Decima Research to quantify how the downloading of music files through P2P networks influences music purchasing in Canada.

In the second category, researchers mainly measure and examine topological properties of P2P networks [58] by monitoring P2P control packets, as they are small sized and with rich information about connections inside the networks. In the study of [58], the topology of the Napster and Gnutella networks were explored. The detailed measurements of P2P networks are performed with crawler and probers. Sen et al. [59] analyzed flow-level data from a large ISP to unleash the dynamic of P2P networks. Bhagwan et al. [150] recently have studied the availability of Overnet by probing the crawled hosts and proved that the

IP based measurement in previous studies had dramatically under estimated the number of peers. In [151], Qiao et al. compared Gnutella and Overnet, and figured out they are on a similar dynamic level. Unfortunately, all of these researches only focus on a subset of peers in networks.

Privacy and anonymity have become significant issues in P2P networks as more and more efforts are applied to P2P networks by copyright holders, such as RIAA and MPAA.

People try to hide file transfers and their identities by various means (e.g., use of proxy servers) in P2P networks. These efforts actually cannot ensure complete anonymity.

Although the pioneering Freenet network [148] and Mnet [149] offer true anonymous

P2P networks, they are not efficient or popular in file sharing.

143

Currently there is very few research study focused on developing a systematic approach to collect, track people’s activities on P2P networks, e.g. how they use different P2P applications or clients to download or upload copyrighted contents, and create a high quality data collection to discover the download patterns. Additionally, the collection can easily help copyright owners to identify the details of infringements and send out DMCA notices.

5.2 Implementation of Building ASKE Data Collection

There are three key components in the ASKE framework to build high quality data collections: Resource Identifier, Spider Agents, and Content Filter. Unlike the web documents, P2P data cannot be collected with the web crawlers or spiders. In this section, we describe the details of implementation about these three components for eDonkey2000, Gnutella, and BitTorrent.

5.2.1 Resource Identifier

Given a copyrighted material, the resource identifier is responsible for looking for the resources to download the given material, which usually include title, time, artist, and copyright owner information. There are mainly two ways in P2P networks to identify resources. The first way is that no matter what copyright contents are tracked, we collect any resources in P2P networks; the other way is based on the given title, time, artist, and copyright owner information, to use the search functions provided in P2P networks to find the related resources. To get a complete cover range, we implement both methodologies in eDonkey2000, Gnutella, and BitTorrent.

1. eDonkey2000

There are a lot of ed2k websites to publish ed2k links for movies, tv shows, music etc. on the Internet, such as VeryCD ( http://www.verycd.com ) and ShareReactor

144

(http://sharereactor.com ). Also there are some other web sites providing the search functions for ed2k links, e.g. FileDonkey ( http://filedonkey.com/ ) and Figator (http://figator.com/). The systematic process to locate and archive ed2k links from web sites are described as follows:

• Compile a list of titles that are popular or concerned by copyright owners • Identify websites that post and update ed2k links frequently: compile a list of high quality web sites with domain experts’ help; use popular pirated titles to search on general search engines to locate this kind of websites. • Crawl ed2k links: build an automatic web crawler to grab the ed2k links from the web sites identified in last step periodically. • Search on ed2k link search engines: with the list of titles, we can search on ed2k link search engines to identify the related ed2k links periodically. • Archive ed2k links: since all ed2k links are in similar formats, it’s easy for us to store them in a database with file name, file size, and hash codes; here we use hash codes as distinct index to identify ed2k links uniquely, as some ed2k links may use different file names with same hash codes.

The process to collect ed2k links in ed2k networks is a little bit different.

• Build fully compatible ed2k resource agents: since we need to search ed2k networks directly, an agent that can connect and search to ed2k networks has been built. • Maintain a list of ed2k servers: ed2k servers manage the information distribution and work as several central dictionaries which hold the information about the shared files and their respective client locations; so a list of high quality and popular ed2k servers has to be built and maintained, given ed2k servers are got shutdown very frequently. • Connect to ed2k servers: to be more efficient, we start tens of ed2k resource agents to connect multiple ed2k servers at the same time. • Search on ed2k networks: each ed2k server will return ed2k links matched to titles and corresponding content types; the main issue here is that most ed2k servers have self protection mechanism, meaning when requests from same IP address hit the server too often, it will automatically reject the connection from that IP address, in this case, we have to control the ed2k resource agents to search ed2k servers with some random delays. • Archive ed2k links: this step is same as above.

2. Gnutella

145

The resource identifier for Gnutella is kind of simple, since in Gnutella Networks, there is no central server. The Gnutella resource agents we built connect to as many super nodes as they can and send out search requests containing title keywords and content types.

3. BitTorrent

In BitTorrent networks, contents are distributed via torrents, which are small files around few kilobytes and contain file names, their sizes, trackers, and checksum info etc. The BitTorrent resource identifier focuses on how to collect torrents on the Internet, parse them, and store them into databases.

The process to locate and archive torrent files is similar to ed2k, as there are huge amount of web sites to publish torrents in different categories, such as ThePirateBay (http://thepiratebay.org ) and Monova ( http://www.monova.org ). More torrent search engines are available also, including isoHunt ( http://isohunt.com ) and Torrentz (http://www.torrentz.com ). The behavior of the BitTorrent resource identifier is similar to the one for ed2k:

• Use title keywords to search general search engines to identify torrent URLs • Maintain a list of torrent web sites and torrent search engines • Grab torrent URLs from torrent web sites • Search title keywords in torrent search engines and get torrent URLs • Download torrents via the torrent URLs identified from the steps above • Parse torrents to get file names, file sizes, trackers, and hash codes which are the unique identifier for torrents

5.2.2 Spider agents

The perfect spider agents would be complete and instantaneous for any given resources

(ed2k links or torrents). However, in practice such perfect spider agents don’t exist as the following conditions cannot be met:

• Rapidly Changing Topology: spider agents target to create a perfect and complete

snapshot of the P2P network topology in a very short time delay. However,

146

spiders actually contact peers progressively. Therefore, the connection speed of

the spider and its limited hardware resources would make capturing a complete

snapshot take a lot of time. The more time the spider contact peers, the more

distorted the captured snapshot is compared to the instantaneous ideal as more

peers join or leave during the spidering process.

• Unreachable Peers: the snapshots collected by spiders are often incomplete

because there are always unreachable peers, which are either located behind a

firewall or overloaded. We can make the spiders more persistent with longer

timeouts when connecting peers, but in the other hand, it will increase the spider

duration and thus generate distortion.

There is no perfect solution to solve these two challenges. The way to mitigate the problems is to develop fast spider agents and run as many spider agents as we can concurrently. In this section, we discuss how to develop fast and efficient spider agents that can capture snapshots for P2P networks. Figure 5.1 presents the high level architecture of P2P spider agents.

Our P2P spider agents mainly deploy the following features to collect peers’ activities, as described below.

1. Distributed Architecture

In order to archive a high degree of concurrency and to effectively utilize available resources on multiple servers, the spider agents employ a three-level master-slave architecture, described as in Figure 5.1.

147

The first level is Top Master Process (TMP), which is responsible to extract resources

(ed2k links, torrents) from the resources database, and dispatch the resources to the second level Sub Master Process (SMP). The SMP on the second level connect to P2P networks, and collect peers that associate to the resource which is uniquely identified by a hash code. The Sub Master Process schedules multiple slave processes on the third level that act as virtually independent spiders and communicate with the peers in parallel.

Each slave has an independent queue of finite number elements to contact, which are filled by SMP. An element in the queue includes the peer information and the resource identifier (usually are hash codes). The whole procedure would never stop, since TMP will keep retrieving from the resources database recursively.

Figure 5.1 The Architecture of P2P Spider Agents

148

2. Asynchronous Communications

Each slave process reaches hundreds of peers in parallel to get peer information asynchronously. To make sure all salve processes are busy but not overloaded, we implement an adaptive schedule algorithm based on each slave node’s resources. Based on the reported resource usage for each slave, the number of peers that are assigned to each slave is adjusted accordingly.

Each slave maintains a threshold, call max_slaves , which makes sure that there would be no new connections when the maximum number of open connections is achieved.

3. Appropriate Timeouts

Many peers are unreachable because of firewalls or overloaded. As we described before, we need to find appropriate timeouts because of the tradeoff when increase and decrease timeout values for spidering. If we set low timeouts, e.g., less than 10 seconds, we will see a lot of unreachable peers; in the other hand, if we set high timeouts, e.g., 5 minutes, the spider agents run with very low efficiency. In our system, we set the timeout to 30 seconds, which is appropriate for our P2P applications. For the peers that are overloaded, we can always try to connect in next round.

4. Utilization of P2P client features

Some protocol features would help a lot to promote spider agents’ efficiency. For example, Gnutella clients implement a two-tiered network structure with ultrapeers and leaf peers. Most of the peers are leaves that are connected to several ultrapeers. So that

149

when connecting to an ultrapeer, spider agents explore all the leaf peers that are connected to the ultrapeer. By this approach, we are able to capture all the nodes and links by connecting a relatively small number of peers. In addition, in most popular

Gnutella clients, such as LimeWire, they allow a quick query to a peer to receive details of peer information, thus spider agents can learn the addresses of the peer’s neighbors, and for an ultrapeer, the addresses of its leaves.

5.2.3 Content Filter

The noise or garbage data exists in the whole process of building data collection. When we collect resources for a movie title “Star Trek”, the resource identifier may mistakenly collect torrents for “Star Trek” TV shows. Since P2P networks are unmanaged, anyone could create fake ed2k links or torrents that actually contain garbage data, for example a file with all 0 filled, and named as “avatar.3d.2009.blu-ray.mkv”, or copyright owners may poison P2P networks for their copyright contents. Also some bad guys would spoof

IP addresses with bogus IPs that actually don’t exist. Content Filter is responsible to rate resources to correct titles, clean up fake resources, and identify bogus IPs.

1. Rate names

To rate a filename or a folder name to its corresponding title correctly, we have to understand the naming rules in P2P networks. Here we use movie filenames and folder names as the example. The following is a typical filename of a movie release:

Title.of.the.Movie. YEAR .Source .[ Language ]. Codec -GROUP[ .Filetype]

150

• YEAR: the release year of the movie

• Source: it’s pirated movie release type with respective sources, ranging from the

lowest quality to the highest, including Cam (CAM), Telesync (TS, TELESYNC,

PDVD), Workprint (WP), Screener (SCREENER, DVDSCR, DVDSCREENER,

BDSCR), R5 (R5), Telecine (TELECINE, TC) , DVD-RIP (DVDRip) , DVDR

(DVDR), HDTV (TVRip, DSR, STV, PDTV, HDTV, DVBRip), BD/BR Rip

(BDRip, 1080p.Blu-Ray, 720p.Blu-Ray, Blu-Ray)

• Language: this field is optional, it specifies the language in the movie, e.g.

English, Spanish, Italian, etc.

• Codec: it describes what video standard this release follows, such as VCD, SVCD,

DivX, Xvid, x264, h264 etc.

• Group: it identifies the group name who compresses this movie and releases it on

P2P networks.

• Filetype: the file type of the file, such as avi, mkv, mp4, mpg, and rmvb etc.

A typical movie release will look like:

Garfield.A.Tail.Of.Two.Kitties.2006.720p.BluRay.x264-AVS720

Here “Garfield A Tail Of Two Kitties” is the movie title; the theatre date of the movie is

2006, the source of this release is Bluray; the resolution of the release is 720p, which means 1280x720; the codec is x264; the release group is called AVS720.

With the understanding of naming rules, we can build a rating engine to map resources to titles:

151

• Compile a complete title list of a given copyrighted material: for example, the

movie “Garfield A Tail Of Two Kitties” has another name “Garfield II” or

“Garfield 2”.

• Extract title and year info from filenames or foler names: remove other

information about the file or folder

• Process special characters: like “.”, “-”

• Calculate the similarity between the processed file name or folder name with the

given title: if the similarity is higher than 90%, then we believe this resource is for

this title

2. Remove fake resources

We apply two methods here to remove fake resources.

• 0day check: it is based on an assumption, “0day releases are always the first

releases in P2P networks”. 0day refers to any copyrighted work that has been

released the same day as the original product, or sometimes even before. It was

considered a mark of skill among warez distro groups to crack and distribute a

program on the same day of its commercial release. We collect all 0day releases

details (names, groups, dates) from VCD Quality ( http://www.vcdq.com ) and

NfoDB ( http://nfodb.com ). Any resources that we find in P2P networks and are

earlier than the first 0day release are rated as “FAKE”

• Chuck check: this methodology is based on the empirical research. A lot of fake

files on P2P networks are actually filled with all zeros. We use spider agents to

152

download chunks of the file and check if the chucks are all zeros. Though this

method is a little bit slow, it’s useful when the first method is not applicable, e.g.

0day release has shown up.

3. Identify bogus IPs

When spider agents retrieve IP address list from ed2k servers, BT trackers, or Gnutella ultrapeers, the list may have been contaminated by bogus/spoofed IPs. Instead of handing the IP list directly to slave processes to check, we need to filter out these bogus IPs first for efficiency consideration. A lot of the bogus IPs are actually unallocated. It’s easy for us to filter them out by comparing them against the bogon list from bluetack

(http://bluetack.co.uk/ ).

5.3 Services and Case Study

With the supports from efficient spider agents, we can provide copyright owners multiples services that help them understand how, when, and where their materials is being pirated online as a means of protection, enforcement, and comparative analysis to more effectively and efficiently plan for the present and the future. In this section, we present how users can use KFC to define their profiles and reports. In addition, a case study on the movie Watchmen is reported.

5.3.1 Services for Copyright Owners

153

To help copyright owners understand the whole picture of their content piracy in P2P networks, we create an Anti-Piracy system. Users can login the site and customize their profiles. KFC in the

ASKE framework is actually the back end to the customized services. As the data we collected are stored in structured databases, the search function is kind of common to just query the databases.

1. Customized Profile

Figure 5.2 presents the interface for users to customize their profiles. Below are the settings users can adjust for a better user experience and for displaying only the information you would like to see when navigating through the system.

• Location: This setting allows users to choose the country or region in which users are

currently located.

• Language: This setting will customize the language displayed in the interface based on

the language chosen. The interface supports all languages but is dependent on the

language packs installed in users’ browsers.

• Date/Time Format: This lets users chose US, European or internet ISO date

representation format.

• GMT setting: The GMT setting lets users chose your local time zone so that when time

related information is displayed the interface or reports will calculate users’ time zone.

• SMTP server: This setting controls from where the system sends the email messages and

reports.

• From Field: This field represents the email address that will be displayed to the recipient

of a report that has been sent out. If someone selects to use the reply function in their

email client this is where the reply would be routed to.

154

• Reply to Field: This works the same as the last section does. However this is also used

by the different email application for rules and filtering.

• My overview settings: These settings let users choose how to customize the overview

page in the system and in reports. It allows users to choose the columns of the 3 sections

on the page and select or deselect columns to display.

• Interested titles: This function provides an entry for users to upload the details of titles

they are interested in to follow up in P2P networks. The input file can be an excel file or

an csv file in pre-defined format.

Figure 5.2 Customize user profile 2. Overview Page

155

The overview page provides users the quick snapshot of targets and pirates as Figure 5.3, such as Today’s Top Pirates, Monitoring, and Fresh Title Releases. Each entry has a magnifying glass for users to check the detail record of the target, such as time, IP address, port, infringed title, infringed file name, hash codes etc.

Figure 5.3 Infringement overview page 3. Search Interface

Users can search the details of piracy via IP, Zipcode, title etc. Search functions here are used to reveal relations and associations between targets on different protocols.

Figure 5.4 Infringement search interface 4. Analysis Report

156

With solid piracy data stored in the database, the Anti-Piracy system can generate reports with business intelligence easily. For example, we can figure out the top 5 infringing countries based on trackers count and peers count during a given period for a movie title, see Figure 5.5.

Figure 5.5 Top 5 countries of peers and trackers for “Quantum of Solace” between 11/03/2008 to 11/21/2008

157

5.3.2 Case Study – Watchmen

To illustrate the usage of the ASKE framework applied in P2P world, we conducted a case study on a movie Watchmen for a month from 3/6/2009 to 4/5/2009 to generate a high-level overview of global P2P piracy activity. The main P2P protocols we monitored here are eDonkey and BitTorrent.

Watchmen premiered on March 4, 2009, to a worldwide audience filled with tremendous hype and anticipation. In the United States, it was the all-time widest-release of an R rated film. As of April 13, 2009, it is the #1 highest-grossing R rated film of 2009. It is also the #1 highest-grossing Alan Moore adaptation. Such a ground-breaking film is bound to be a catalyst for significant online piracy.

Watchmen was first made available online on Friday, March 6, 2009. Since then, the film has been downloaded over 730,000 times in 213 countries by approximately 627,000 distinct individuals. It has been uploaded over 795,000 times from 16 countries by approximately 92 servers.

The most popular versions of the film released across all protocols were in English and were telesync versions of the film. The quality is quite inferior relative to a DVD. Piracy surrounding the film is expected to surge to much higher levels as soon as better quality versions become available. Appendix F list the resources (ED2K links and BitTorrent files) we located for Watchmen.

158

Figure 5.6 presents the distribution of piracy among different protocols. BitTorrent is by and large the most popular protocol available for sharing content online. 91% of total infringements for Watchmen occurred on the BitTorrent protocol, with 9% occurring on eDonkey. This is common for film titles because of the size of files involved. BitTorrent is the ideal protocol for sharing large files because it does not cause network strain, nor does it require exhaustive amounts of bandwidth on one particular server or network. By spreading the network load, facilitators (seeders), peers and trackers can easily communicate and share content with minimal resources.

Figure 5.6 Protocols breakdown In Figure 5.7, we present the daily trend of infringements. The overall trend for total infringements of Watchmen is clearly downwards. This is common for a film which does not have high-quality pirated versions available. The most piracy occurs within one week

159

of general release; especially when there is not a tiered release schedule beyond a few days.

In this case study, we also identified the top 10 ISPs with most peers which were pirated the movie. Telecom Italia is the leading provider of broadband service in Italy, and singlehandedly represents the majority of peer infringements worldwide originating from one ISP. Telefonia de Espana had slightly less infringements. Open Hosting Limited was split approximately 60% to 40% in Germany and the United Kingdom. Deutche

Telekom AG is based in Germany. SBC Internet Services and Comcast Cable, and Road

Runner are the leading ISPs in the United States. France Telecom had the sixth-most and represents peer infringements originating from France, Algeria, French Guiana,

Guadeloupe, Madagascar, Martinique, Niger, and Reunion. British Telecommunications is based in the United Kingdom, and Polish Telecom is based in Poland.

The top five releases were in English, followed by an Italian version, and two German versions. These releases were primarily telesync versions of the film, with cam versions available as well. During the testing period, there were no high-quality pirated releases of the film. In total, there were over 150 pirated versions of film available worldwide.

160

Figure 5.7 Daily infringements trend – all infringements

Figure 5.8 Top 10 ISPs

161

Figure 5.9 Top 10 filenames

5.4 Conclusions

In this chapter, we mainly discussed how to monitor P2P networks using ASKE framework. We reviewed different P2P protocols and current research topics in P2P networks. The detail of implementation of building data collection is presented. Also we reported the Anti-Piracy system which uses the P2P data collection to serve users with customized profiles. At last, we presented our case study results for monitoring the movie

Watchmen piracy on eDonkey network and BitTorrent Network.

162

CHAPTER 6

CONCLUSIONS AND FUTURE DIRECTIONS

The Internet provides the largest knowledge repository across domains, applications, and countries. It is desirable for researchers, managers, and government agencies to access, analyze and share such huge information. This dissertation proposes an application specific framework to support Internet searching, browsing, and analysis with effective and efficient approaches. This chapter summarizes the main conclusions and contributions of this dissertation, and suggests future directions.

6.1 Conclusions

In this dissertation, we explore an effective ASKE framework to build structured and semantic data repositories, and support keyword search and semantic search. The framework is consistent with the architecture of most search engines. It enhances three extensions to this basic structure: various data retrieval ability; semantic data support; and post-retrieval analysis. Various techniques and algorithms that could facilitate knowledge discovery are applied in the framework.

By reviewing the characteristics of different data on the Web, we recognize that we have to figure non-traditional ways to approach the problems that are resulted by these features.

In Chapter 2, we explore unstructured and structure data, data on online forum and social networking sites, and P2P data.

163

To see how the framework can be applied into specific domains, we describe how to develop an experimental Web-based counterterrorism knowledge portal, called the Dark

Web Portal, to support the discovery and analysis of Dark Web information and provide an intelligent, reliable, interactive, and convenient interface with for the counterterrorism experts. Systematic approaches to identify resources, collect different types of Dark Web data, and build semantic data collection are presented in details.

Monitoring P2P data on the Internet is another study with ASKE framework. We describe how to build an anti-piracy system, which generates insights of P2P download activities and provides an effective anti-piracy strategy to combat piracy in P2P networks. We propose a feasible approach to implement the components in ASKE for different P2P protocols, and report the challenges and problems during the process we implement them.

The anti-piracy system provides customized service to users via KFC. A case study focused on the movie “Watchmen” is discussed with details based on the data collection we built via the ASKE framework.

6.2 Future Directions

I plan to expand my research in several directions. First, I plan to continue development of effective and efficient techniques and algorithms that support building semantic repositories and semantic search, in particular, the implementation of Metadata Extractor.

Secondly, I plan to conduct large scale interactive user evaluations of such application specific Web portals. This is an area has not been investigated. I am also interested in

164

continuous investigation and measurement of P2P networks, as the growth and changes keeps happening in P2P world. In addition, I will experiment the framework in more application domains, such as some scientific research domain, e.g. intelligent transportation systems.

165

APPENDIX A

US DOMESTIC EXTREMIST GROUPS AND URLS

"1" represents that the group name is listed at the specific resource. For "White Supremacy", only SPLC listed different branches of this group, while other resources use KKK as a general group name. For consistency, we add frequency count to all the branches listed accordingly, assuming all branches listed by SPLC are considered part of the bigger "KKK" group by other resources too.

Acronyms:

ADL - Anti-Defamation League

FBI - Federal Bureau of Investigation

SPLC - (Southern Poverty Law Center)

Black Separatist ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory United Nuwaubian Nation of Moors 1 1 http://www.geocities.com/Area51/ Corridor/4978/unnm.html New Black Panther Party 1 1 2 http://hamp.hampshire.edu/~cmnF 93/panthers.html Nation of Islam 1 1 1 3 http://noi.org/ Christian Identity ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory America's Promise Ministries 1 1 http://amprom.org Artisan Publishers 1 1 http://www.artisanpublishers.com

166

Aryan Covenant Church/ACC Services 1 1 By YahwehÕs Design 1 1 Church of Christ in Israel 1 1 Fellowship of GodÕs Covenant People 1 1 Gospel Broadcasting Association 1 1 Upper Room Identity Fellowship 1 1 Virginia Christian Israelites 1 1 Dogma of Christian Identity 1 1 http://www.geocities.com/onemans mind/ Rick Ross 1 1 http://www.rickross.com/ Cry Aloud Cybermagazine 1 1 http://www.cryaloud.com/ Harp Of David Ministries 1 1 http://harpofdavid.homestead.com/ CELTIC3.html First Century Christian Ministries 1 1 http://www.secondexodus.com/ html/russford/firstcenturychristian ministries.htm Christian Research 1 1 2 http://www.equip.org/ Christian Separatist Church Society 1 1 2 http://www.christianseparatist.org/ Church of the Sons of Yhvh 1 1 2 http://www.aryannationsknights.co m/socy2.htm http://www.churchofthesonsofyhvh .org/ Church of True Israel 1 1 2 http://www.churchoftrueisrael.com / Covenant Church of Yahweh 1 1 2 http://www.nccg.org Crusade for Christ 1 1 2 http://www.ccci.org/ Gospel Ministries 1 1 2 http://www.gospelfortoday.org/ House of Yahweh 1 1 2 http://www.yahweh.com/ Kinsman Redeemer Ministries 1 1 2 http://www.kinsmanredeemer.com/ Lord's Work 1 1 2 http://www.freespeech.org/thelords work/ Mission to Israel 1 1 2 http://www.missiontoisrael.org/ New Covenant Bible Church 1 1 2 http://www.covenantbiblecollege.a b.ca/ Virginia Publishing Company 1 1 2 http://www.wordnews.com/ Weisman Publications 1 1 2 http://www.seek-info.com/

167

Westboro Baptist Church 1 1 2 http://www.godhatesfags.com/ Christian Legal Reformation Club 1 1 2 http://www.clrc.net/ http://www.clrc.net/clrc.html Northwest Kinsmen 1 1 2 http://www.cris.com/~nwk/nwk.ht m Christian Separatist Church Society 1 1 2 http://www.christianseparatist.org/ Mission to Israel 1 1 2 http://www.missiontoisrael.org/ Church of Jesus Christ 1 1 1 3 http://thechurchofjesuschrist.com/ Scriptures for America Ministries 1 1 1 3 http://www.scripturesforamerica.or g http://www.christianidentity.org/ho me.htm Kingdom Identity Ministries 1 1 1 1 4 http://kingidentity.com/ Militia ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory 42nd/58th/59th Brigades Missouri 1 1 http://www.geocities.com/CapitolH Militia ill/4274/ The Ohio Unorganized Militia 1 1 http://www.oumaac.com Assistance And Advisory Committee The Viper Reserves 1 1 http://www.geocities.com/CapitolH ill/Lobby/8786/viper.htm Minnesota Minutemen Militia 1 1 http://www.geocities.com/CapitolH ill/Lobby/6745/default.html Republic of Texas Defense Force 1 1 http://web2.airmail.net/reptex1/def ault.htm Central Ohio Unorganized Militia 1 1 http://www.infinet.com/~pandar/ California Militia 1 1 http://pw1.netcom.com/~stevep/we lcome.html North Carolina Citizen Militia 1 1 http://www.netpath.net/~jeffr/nccm .htm Michigan Militia Home Page 1 1 http://militia.gen.mi.us/ Militia of Montana 1 1 http://www.nidlink.com/~bobhard/ mom.html The Constitution Society 1 1 http://www.constitution.org Louisiana Unorganized Militia 1 1 http://www.orion- cs.com/freedomforum/

168

Southeastern Ohio Defense Force 1 1 http://members.aol.com/RMORGA N762/default.html 7th Missouri Militia 1 1 http://www.mo- net.com/~mlindste/7momilit.html Southeastern States Alliance Militia 1 1 http://www.geocities.com/CapitolH Headquarters ill/9852/saa.htm United States Theatre Command 1 1 http://www.bignet.net/~eagleflt/ The Real Deal 1 1 http://www.sfol.com/sfol/santafe/re aldeal/default.html Pomona Valley Militia 1 1 http://home.earthlink.net/~wsranch /webdocs/Pomona_Valley_Militia. htm The 2nd Cavalry - 14th Division 1 1 http://www.geocities.com/CapitolH Kentucky State Militia page ill/Lobby/9659/default.html The Palmetto State Guard 1 1 http://www.geocities.com/CapitolH ill/Senate/4952/default.html Militia Channel Homepage on Efnet 1 1 http://www.mdc.net/~militia/ Ted Davis Memorial Brigade 1 1 http://www.ocnsignal.com/y2k- militia1.htm Montana Unorganized Militias 1 1 http://www.angelfire.com/mt/militi awar/militia.default.html Militia Report - Columbiana County - 1 1 http://hometown.aol.com/locowrpo Ohio ny/ 508th Airborne Regiment - Texas 1 1 http://welcome.to/508thAirborne National Defense North Mississippi Militia 1 1 http://members.xoom.com/nm_mili tia/index2.htm Arizona Unorganized Militia 1 1 http://www.american- unlimited.com/arizonauno.htm The Guardians 1 1 http://www.guardians.org/ Ramblings 1 1 http://www.coolmedia.net/cbg/ram ble/ramblings.html South East Texas Defense Force 1 1 http://www.geocities.com/Pentago n/8410/setdf.html Texas Shock Front Elite 1 1 http://members.xoom.com/TexasS FE/

169

New Mexico Citizens Regulated Militia 1 1 http://www.nmex.com/militia/milit ia.htm Wayne County Militia of Michigan 1 1 http://www.geocities.com/CapitolH ill/1392/index2.htm South Carolina Militia Corps 1 1 http://pw1.netcom.com/~dan3/scm c.htm Thirteenth Texas Infantry Regiment 1 1 http://freeweb.pdq.net/metalryder/1 3thTIR/ Iowa Unorganized Militia 1 1 http://www.angelfire.com/ia/IUM/ Militia of Montana 1 1 http://www.montana.com/militiaof montana/ Virginia Citizen’s Militia 1 1 http://vcm.freeservers.com/ Viper Militia 1 1 http://www.vipermilitia.org/ Washington County (Maine) 1 1 http://www.angelfire.com/me2/Con Constitutional Militia stitutionalMilita/default.html Hudson Highlands Free Militia 1 1 http://www.angelfire.com/ny2/hhf m/ California Militia Training Center 1 1 http://home.earthlink.net/~jbk97/ Danville Militia 1 1 http://members.theglobe.com/DEM M/file_name.html Pennsylvania State Military Reserve 1 1 http://www.navpoint.com/~pasmr/ Pomona Valley Militia 1 1 http://home.earthlink.net/~wsranch /webdocs/Pomona_Valley_Militia. htm South Dakota Unorganized Militia 1 1 http://www.geocities.com/CapitolH ill/Congress/7011/ Marietta Militia 1 1 http://mariettapa.com/marietta_mili tia.html Michigan Militia of Wayne County 1 1 http://www.michiganmilitia.org/ New Jersey Militia 1 1 http://www.exit109.com/~njm/defa ult.htm Militia of North Dakota 1 1 http://community- 1.webtv.net/max722/MILITIAofN ORTHDAKOTA/ Minnesota Minuteman Militia - Fifth 1 1 http://members.xoom.com/mmmst Brigade _cloud/

170

Missouri 51st Militia 1 1 http://www.mo51st.org Michigan Militia - Tenth Brigade 1 1 http://www.geocities.com/~cowbo y140/ The Intelligence Report 1 1 http://intelreport.freeservers.com/ California Militia 1 1 http://www.geocities.com/CapitolH ill/Congress/2608/welcome.html The Official Pack 44 Militia 1 1 http://marina.fortunecity.com/long mark/311/ The Patriot Underground 1 1 http://www.geocities.com/CapitolH ill/Parliament/1691/ New Mexico’s Liberty Corps 3rd 1 1 http://www.users.uswest.net/~toad Brigade 419/ New Mexico Militia 1 1 http://www.zianet.com/toad419/ Militia of Florida 1 1 http://www.militia-of-florida.com/ Michigan Militia 1 1 http://www.michiganmilitia.com/ The People’ss Militia 1 1 http://www.angelfire.com/il/fear/ The Southern Indiana Regional Militia 1 1 http://www.fortunecity.com/victori an/crayon/881/default.html/ The Citizens Militia of Maryland 1 1 http://www.expage.com/page/citize nsmilitiaofmaryland So. Cal. High Desert Militia 1 1 http://www.freeyellow.com/membe rs8/highdesertmilitia/default.html New Mexico Militia 1 1 http://homes.acmecity.com/rosie/s miley/145/default.html Militia of East Tennessee 3rd Brigade 1 1 http://www.geocities.com/met3rdbr igade/ Sons of Liberty Militia 1 1 http://community- 1.webtv.net/We_The_People_/Son sofLibertyMilitia/ United States Special Operations 1 1 http://hometown.aol.com/USSOC Citizens Militia of Florida M/page3/default.htm Connecticut 51st Militia 1 1 http://expage.com/page/ctmilitia Michigan Militia - 4th Division - 19th 1 1 http://www.geocities.com/Pentago Brigade - Roscommon County n/Bunker/6210/default.html Midsouth Liberty Alliance 1 1 http://members.xoom.com/nm_mili tia/

171

508th Regiment 1 1 http://randymiller.webjump.com/ Neo-Nazis ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory Aryan Nations COJCC 1 1 Aryan Nations-offshoot 1 1 Aryan Werwulfe Brotherhood 1 1 German American Nationalist PAC 1 1 Inter-National Socialist Party 1 1 Liberty Bell Publications 1 1 http://www.lbp2.com/ National Socialist German Workers 1 1 Party SS Regalia 1 1 Nazi Low Riders 1 1 Women for Aryan Unity 1 1 http://www.rac-usa.org/wau/ http://www.creator.org/womenforu nity/ http://members.odinsrage.com/wau / http://www.wauhqs.cjb.net/ http://www.wcotc.com/womenforu nity/ http://www.front14.org/wau http://www.stormfront.org/crusader /texts/wau/ http://www.crusader.net/texts/wau/ http://www.natall.com/aryan- page/wau/wau_index.html The Library 1 1 http://ourhero.com/index2.html Aryan Renaissance Society 1 1 2 http://members.odinsrage.com/ars/ National Socialist Vanguard 1 1 2 http://www.alpha.org/nsv/ SS Enterprise 1 1 2 http://ssenterprise.tripod.com/book shop.htm White Revolution 1 1 2 http://www.whiterevolution.com/ Knights of Freedom - Utah State 1 1 2 http://www.io.com/~jack88/ American Nazi Party 1 1 1 3 http://www.americannaziparty.com /

172

National Alliance 1 1 1 3 http://www.natall.com/ National Socialist Movement 1 1 1 3 http://www.nsm88.com New Order 1 1 1 3 http://theneworder88.com/ Christian Defense League 1 1 1 1 4 http://www.cdlreport.com Sons of Liberty 1 1 1 1 4 http://www.geocities.com/sons_of_ liberty_mgs/index.html White Aryan Resistance 1 1 1 1 4 http://www.resist.com/ World Church of the Creator 1 1 1 1 1 5 http://www.anti-semitism.net http://www.creator.org Aryan Nations 1 1 1 1 1 1 6 http://twelvearyannations.com http://www.aryan-nations.org/ Neo-Confederate ADL FBI SPLC Web Hate Total URL Directory Directory The Confederate Society of America 1 1 http://www.deovindice.org South Carolina League of the South 1 1 http://www.palmetto.org/ The Southern Party 1 1 http://www.southernparty.org/ Council of Conservative Citizens 1 1 2 http://www.cofcc.org/ the League of the South 1 1 2 http://www.dixienet.org/ Southern Independence 1 1 2 http://members.aol.com/GrayFox6 5/south.html League of the South 1 1 1 3 http://www.dixienet.org Racist Skinhead ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory Confederate Hammerskins 1 1 Eastern Hammer Skins 1 1 Empire State Skinheads 1 1 National Racist Skinhead Front 1 1 Northern Hammerskins 1 1 Western Hammerskins 1 1 Celtic Knights 1 1 2 http://front14.org/celticknights/ Keystone State Skinheads 1 1 2 http://www.keystonestateskinheads .com/ Las Vegas Skinheads 1 1 2 http://24.234.43.14 http://www.lasvegasskinheads.net Midland Hammerskins 1 1 2 http://www.midlandhammerskins.c om/

173

National Skinhead Front 1 1 2 http://www.whitevictory.com/ Outlaw Hammerskins 1 1 2 http://www.outlawhammerskins.co m/ Hammerskin Nation 1 1 2 http://home.att.net/~wpsh8814/ American Front 1 1 2 http://www.thirdposition@america nfront.com/index2.htm http://www.americanfront.com http://web2.airmail.net/bootboy/af. htm http://www.geocities.com/CapitolH ill/3343 White Supremacy ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory Liberty or Death 1 1 http://www.geocities.com/CapitolH ill/Lobby/2945/ Fourteen Words Press 1 1 http://www.14words.com/ American Coalition of Third Positionists 1 1 http://3rd.org/ Heritage Lost Ministries 1 1 http://heritagelost.org/frame/ Voice of Citizens Together/American 1 1 2 http://www.americanpatrol.org/ Patrol I Love White Folks 1 1 2 http://www.ilovewhitefolks.com/ MOTM(Mothers of the Movement) 1 1 2 http://www.sigrdrifa.com/motm/ Templar Knights of the Ku Klux Klan 1 1 1 0 3 White Knights of the Sovereign State of 1 1 1 3 Missisippi Crosstar - The Nationalist Movement 1 1 1 3 http://www.nationalist.org/default. html White Aryan Resistance 1 1 1 3 http://www.resist.com/ http://www.tommetzger.net/ http://www.tommetzger.org/ http://www.racehate.com/ http://www.free.cts.com/crash/m/m etzger/ Confederate Knights of the Ku Klux 1 1 1 1 4 http://www.bayouknights.org/confe Klan derateknights.htm Divine Knights of the Ku Klux Klan 1 1 1 1 4

174

Georgia White Knights of the KKK 1 1 1 1 4 Invisible Empire White Knights of the 1 1 1 1 4 http://iewk.kukluxklan.cc KKK http://konfederationklans.org/mem bers/iewk/ Knight Riders of the Ku Klux Klan 1 1 1 1 4 http://home.beseen.com/politics/aki a88/ Knights of Yahweh 1 1 1 1 0 4 http://members.tripod.com/knights ofyahweh/ Royal Confederate Knights of the Ku 1 1 1 1 0 4 Klux Klan South Arkansas Knights 1 1 1 1 0 4 http://www.bayouknights.org/south arkknights.htm Southern Mississippi Knights of the Ku 1 1 1 1 4 Klux Klan Alabama White Knights of the Ku Klux 1 1 1 1 1 5 http://www.kukluxklan.net/ Klan American Knights of the Ku Klux Klan 1 1 1 1 1 5 http://www.awkkkk.org America's Invisible Empire Knights of 1 1 1 1 1 5 http://www.aiekkkk.hostmb.com the KKK http://www.aiekkkk.org/ http://www.aie-usa.com/ http://home.hiwaay.net/~krotos/pic turepage.html Bayou Knights of the Ku Klux Klan 1 1 1 1 1 5 http://www.bayouknights.org/ http://198.69.82.120/PirateWeb/Ba youKnightsKKK/ Brotherhood of Klans 1 1 1 1 1 5 http://www.bokkkkk.net/ http://www.bok.kukluxklan.cc http://members.odinsrage.com/bok kkkk Confederate Crusaders 1 1 1 1 1 5 http://www.bayouknights.org/confe derateknights.htm Free Knights of the Ku Klux Klan 1 1 1 1 1 5 http://www.geocities.com/freeknig htsnc/ Imperial Klans of America 1 1 1 1 1 5 http://www.kkkk.net/ International Keystone Knights of the 1 1 1 1 1 5 http://www.uniqe.com/kkk Ku Klux Klan

175

Invisible Empire Knights of the Ku Klux 1 1 1 1 1 5 http://www.uniqe.com/kkk Klan Knights of the KKK 1 1 1 1 1 5 http://clubs.yahoo.com/clubs/knigh tsofthekkk Knights of the White Kamellia 1 1 1 1 1 5 http://www.kwknational.homestead .com/index.html http://www.angelfire.com/tx5/kwk kkk/ http://members.theglobe.com/klan man1/ http://www.acadian.net/~sandmanh ttp://www.kamellia.com Ku Klux Klan 1 1 1 1 1 5 http://shell.idt.net/~edoneil1/kkkho me.html Mississippi White Knights of the Ku 1 1 1 1 1 5 http://www.mwkkkk.org/ Klux Klan Mystic Knights of the KKK 1 1 1 1 1 5 http://www.mysticknights.org/ http://mwkkkk.europeanwhiteknigh ts.com/ http://mwk.kukluxklan.cc http://akia.yoderanium.com/mwk National Knights of the KKK 1 1 1 1 1 5 http://www.nkkkk.org/ http://www.cnkkk.org http://nationalknights.org http://nkkk.cjb.net/ http://www.geocities.com/national _knights_kkk/ http://www.geocities.comnationalk nightskkk/ http://www.neters.com/web/wwf.sh tml http://akia.cjb.net North Georgia White Knights of the Ku 1 1 1 1 1 5 http://members.surfsouth.com/~ng Klux Klan wk/index.html http://www.theklan.com/ Southern Cross Militant Knights of the 1 1 1 1 1 5 http://militant2.homestead.com/ Ku Klux Klan http://www.homestead.com/militan

176

t/ Southern White Knights of the Ku Klux 1 1 1 1 1 5 http://groups.yahoo.com/group/swk Klan kkk http://www.knightskkk.org http://www.swkkkk.net http://www.swkkkk.com http://www.swkkkk.org http://swk.kukluxklan.cc http://louisiana.cjb.net/ http://southernwhiteknights.homest ead.com/ SS Knights of the Ku Klux Klan 1 1 1 1 1 5 http://www.sskkk.com Texas Knights of the Ku Klux Klan 1 1 1 1 1 5 http://www.texasamericanknights.o rg U S Klans 1 1 1 1 1 5 http://klavalier.freeyellow.com United White Klans 1 1 1 1 1 5 http://expage.com/unitedwhiteklans White Camelia Knights of the Ku Klux 1 1 1 1 1 5 http://www.wckkkk.com Klan Others ADL FBI SPLC Militia Web Hate Total URL Watchdog Directory Directory Carnival Against Capitalism 1 1 Coalition to Save the Preserves 1 1 http://www.earthliberationfront.co m/main.shtml Creativity Movement 1 1 Earth Liberation Front 1 1 http://www.animalliberationfront.c om/ Elohim City 1 1 Greater Ministries International 1 1 Los Macheteros 1 1 http://www.macheteros.com/ New Black Panther Party for Self- 1 1 Defense Reclaim the Streets 1 1 http://www.reclaimthestreets.net/ Southeastern States Alliance 1 1 http://www.geocities.com/CapitolH ill/9852/saa.htm Sovereign Citizens 1 1 Workers' World Party 1 1 http://www.workers.org/

177

United Fascist Union 1 1 http://www.geocities.com/Area51/ Chamber/7344/index.htm American Falangist Party 1 1 http://www.falange.org/ National Patriotic Front 1 1 http://www.geocities.com/Colosseu m/Loge/8461/engl.html American Fascist Movement 1 1 http://www.americanfascistmovem ent.com Deseret National Socialist Syndicate 1 1 http://libreopinion.com/members/d nss/ Folk And Faith 1 1 http://www.folkandfaith.com/ Animal Liberation Front 1 1 2 http://www.animalliberationfront.c om/ Armed Forces for Puerto Rican National 1 1 2 http://www.fas.org/irp/world/para/f Liberation aln.htm Council of Conservative Citizens 1 1 2 Institute for Historical Review 1 1 2 http://www.ihr.org

178

APPENDIX B

INTERNATIONAL TERRORIST GROUPS AND URLS FOR ARABIC GROUPS

Group Name Acro Country Region Detailed Status Source Total nym Region Movimento de la Rwanda Africa Central Active USCFAFL 1 Izquierda Africa Revolucionaria Corsican National FLA Niger Africa Central Tentative USCFAFL 1 Liberation Front- A Africa truce Traditional Wing Kanak Socialist ORA Niger Africa Central Tentative USCFAFL 1 National Liberation Africa truce Front Abkhazia rebels ARL Niger Africa Central Tentative USCFAFL 1 N Africa truce Algetian Wolves FPLS Niger Africa Central Tentative USCFAFL 1 Africa truce Janatha Vimukthi FARF Chad Africa Central Active USCFAFL 1 Peramuna Africa Fatah wing FNT Chad Africa Central Active USCFAFL 1 Africa Forces of Unity MDD Chad Africa Central USCFAFL 1 Africa Legitimate CSNP Chad Africa Central USCFAFL 1 Command D Africa National CNT Chad Africa Central USCFAFL 1 Democratic R Africa Alliance Southern Sudan CNR Chad Africa Central USCFAFL 1 Independence Africa Movement Sudan Alliance FNT Chad Africa Central Active USCFAFL 1 Forces R Africa Sudan People's PDF Chad Africa Central Active USCFAFL 1 Liberation Army Africa Umma Liberation UFD Chad Africa Central Active USCFAFL 1 Army Africa People's FDD Burundi Africa Central Active USCFAFL 1 Democratic Army Africa Popular Front CND Burundi Africa Central Active USCFAFL 1 D Africa Tajik opposition FNL Burundi Africa Central Active USCFAFL 1 Africa Pattani United Frolin Burundi Africa Central Active USCFAFL 1 Liberation a Africa Organization Hizb el Nahda PLPH Burundi Africa Central Active USCFAFL 1 Africa

179

Active Al-Ittihad ADF Uganda Africa Eastern Active USCFAFL, 2 al-Islami Africa US Hizb-ul LRA Uganda Africa Eastern Active USCFAFL, 2 Mujahideen Africa US Red Brigades UNR Uganda Africa Eastern Active USCFAFL 1 F II Africa Japanese Red Army WNB Uganda Africa Eastern Active USCFAFL 1 F Africa Hezbollah- Somalia Africa Eastern Active USCFAFL 1 Gulf/Bahrain Africa Islamic Front for Somalia Africa Eastern Active USCFAFL 1 the Liberation of Africa Bahrain Movement for the SDA Somalia Africa Eastern Active USCFAFL 1 Liberation of Africa Bahrain Shanti Bahini Somalia Africa Eastern Active USCFAFL 1 Africa Fighting SNA Somalia Africa Eastern Active USCFAFL 1 Communist Cells Africa National Liberation SNF Somalia Africa Eastern Active USCFAFL 1 Army - Bolivia Africa Tupac Katari SNM Somalia Africa Eastern Active USCFAFL 1 Guerrilla Army Africa Forces for the Somalia Africa Eastern Active USCFAFL 1 Defense of Africa Democracy National Council Somalia Africa Eastern Active USCFAFL 1 for the Defense of Africa Democracy National Liberation Somalia Africa Eastern Active USCFAFL 1 Forces Africa National Liberation USC Somalia Africa Eastern Active USCFAFL 1 Front Africa Party for the USF Somalia Africa Eastern Active USCFAFL 1 Liberation of the Africa Hutu People Revolutionary REN Mozambi Africa Eastern USCFAFL 1 Subversive Faction AMO que Africa - Commando Unabomber Guatemalan MNR Mozambi Africa Eastern USCFAFL 1 National que Africa Revolutionary Party Chechen rebels Ethiopia Africa Eastern Active USCFAFL 1 Africa South Ossetian AIAI Eritrea Africa Eastern Active USCFAFL, 2 rebels Africa US Interahamwe ELF Eritrea Africa Eastern Active USCFAFL 1 Militia Africa Movement of FRU Djibouti Africa Eastern Active USCFAFL 1 Democratic Forces D Africa of Casamance - Northern Front

180

Movement of FRU Djibouti Africa Eastern Active USCFAFL 1 Democratic Forces D Africa of Casamance - Southern Front National Somali Comoros Africa Eastern Active USCFAFL 1 Congress Africa Palestine Liberation Polisa Western Africa Northern Cease-fire, USCFAFL 1 Front - Abu Abbas rio Sahara Africa often faction broken Mujahedin-e Khalq Zimbabw Africa Southern USCFAFL 1 Organization, e Africa People's Mujahedin Abu Nidal Zambia Africa Southern USCFAFL 1 Organization Africa People's Extra AWB South Africa Southern Active USCFAFL 1 Parliamentary Africa Africa Opposition Revolutionary FDC Angola Africa Southern USCFAFL 1 Communists' Union Africa of Turkey Uganda National FLEC Angola Africa Southern Active USCFAFL 1 Rescue Front II -FAC Africa West Nile Bank FLEC Angola Africa Southern Active USCFAFL 1 Front -R Africa Continuity Army UNIT Angola Africa Southern Active USCFAFL 1 Council A Africa Party of AFR Sierra Africa Western Ousted USCFAFL 1 Democratic C Leone Africa Kampuchea Front de Liberation RUF Sierra Africa Western Cease-fire USCFAFL, 2 du Quebec Leone Africa US Armed Forces for a MFD Senegal Africa Western Active USCFAFL 1 Federal Republic C-FN Africa Chadian National MFD Senegal Africa Western Active USCFAFL 1 Front C-FS Africa Popular FIAA Mali Africa Western Cease-fire, USCFAFL 1 Revolutionary Africa 3/1996 Forces Lorenzo Zelaya Al Faran FPLA Mali Africa Western Cease-fire, USCFAFL 1 Africa 3/1996 Al Hadid MPA Mali Africa Western Cease-fire, USCFAFL 1 Africa 3/1996 Al Jihad ARL Mali Africa Western Cease-fire, USCFAFL 1 A Africa 3/1996 All India Sikh MFU Mali Africa Western Active USCFAFL 1 Students Federation A Africa Khalistan NPFL Liberia Africa Western Disarming USCFAFL 1 Commando Force Africa (slowly) Khalistan ULI Liberia Africa Western cease-fire USCFAFL 1 Liberation Front MO Africa Khalistan Zindabad ULI Liberia Africa Western cease-fire USCFAFL 1 Force MO-J Africa Shan State Army, Guinea- Africa Western USCFAFL 1 or Shan State Bissau Africa

181

Progress Army Brethren MAI Equatorial Africa Western Active USCFAFL 1 (Battalions) of the B Guinea Africa Faithful Internal Opposition Nepal Asia Central USCFAFL, 2 Zviadists Asia US Lahkar-iJhangi Japan Asia Eastern USCFAFL, 3 Asia EU, US Popular Liberation ASG Philippine Asia Eastern Active USCFAFL, 4 Army s Asia UK, AUS, US Ricardo Franco ABB Philippine Asia Eastern Active USCFAFL, 2 Front s Asia US Anjouan Island NPA Philippine Asia Eastern Active USCFAFL, 2 separatists s Asia EU April 19 Movement MILF Philippine Asia Eastern Active USCFAFL 1 s Asia Che Guevara MNL Philippine Asia Eastern Cease-fire USCFAFL 1 Brigade F s Asia Front for the NDF Philippine Asia Eastern Active USCFAFL 1 Restoration of s Asia Unity and Democracy Front for the Philippine Asia Eastern USCFAFL 1 Restoration of s Asia Unity and Democracy - Dini National JRA Japan Asia Eastern Active, USCFAFL, 2 Democratic Front Asia training US terrorists National Sekih Japan Asia Eastern Active USCFAFL 1 Democratic Front otai Asia of Hedayatollah Matin-Daftari National Front Japan Asia Eastern Active USCFAFL 1 Asia National Liberation Japan Asia Eastern USCFAFL 1 Army of Iran Asia (Militant wing of MEK) National Resistance Japan Asia Eastern USCFAFL 1 Movement of Iran Asia Somalia China Asia Eastern Active USCFAFL 1 Democratic Front Asia Somalia Salvation China Asia Eastern Active USCFAFL 1 Democratic Asia Popular Front for FUL Vietnam Asia Southeast Uncertain USCFAFL 1 the Liberation of RO Asia Palestine Hizballah External PUL Thailand Asia Southeast Dormant USCFAFL 1 Security O Asia Organisation Islamic Salvation LTTE Sri Lanka Asia Southeast Active USCFAFL, 3 Front / Islamic Asia UK, US Salvation Army

182

Cabinda JVP Sri Lanka Asia Southeast Active, but USCFAFL 1 Democratic Front Asia limited South Ossetian ARIF Myanmar Asia Southeast Active USCFAFL 1 Rebels Asia Ukrainian Self- KDA Myanmar Asia Southeast Cease-fire USCFAFL 1 Defence Asia Organisation White Legion KIA Myanmar Asia Southeast Cease-fire USCFAFL 1 Asia Anti-Imperialist DKB Myanmar Asia Southeast Active USCFAFL 1 Cell A Asia June 2 KNU/ Myanmar Asia Southeast Active USCFAFL 1 KNL Asia A Red Army Faction KA Myanmar Asia Southeast Active USCFAFL 1 Asia Revolutionary Cells MND Myanmar Asia Southeast Cease-fire USCFAFL 1 AA Asia Anarchist Street NDA Myanmar Asia Southeast Cease-fire USCFAFL 1 Patrol A Asia Children of NDA Myanmar Asia Southeast Cease-fire USCFAFL 1 November Asia Conscientious RSO Myanmar Asia Southeast Active USCFAFL 1 Arsonists Asia Fighting Guerrilla SSA/ Myanmar Asia Southeast Active USCFAFL 1 Formation SSPA Asia Militant Guerilla Myanmar Asia Southeast Active USCFAFL 1 Formation Asia New Group of SUR Myanmar Asia Southeast Active USCFAFL 1 Satanists A Asia Revolutionary UWS Myanmar Asia Southeast Cease-fire USCFAFL 1 Popular Struggle A Asia Fadayan - Minority LLA Laos Asia Southeast Active USCFAFL 1 Faction Asia Freedom LNL Laos Asia Southeast Active USCFAFL 1 Movement of Iran M Asia Iran Liberation ULN Laos Asia Southeast Active USCFAFL 1 Front LF Asia United Front for the GPK, Indonesia Asia Southeast Active; USCFAFL 1 Liberation of FRET Asia talks Liberia ILIN stalled United Front for the Indonesia Asia Southeast Active USCFAFL 1 Liberation of Asia Liberia-Johnson Fighting Islamic OPM Indonesia Asia Southeast Active USCFAFL 1 Group in Libya Asia Islamic Cambodia Asia Southeast Active USCFAFL 1 Renaissance Party Asia Islamic Martyrs BK India Asia Southern Active USCFAFL, 4 Movement Asia EU, UK, JPN Islamic Movement India Asia Southern Active USCFAFL, 2 for Change Asia US Islamic Movement India Asia Southern Active USCFAFL 1 of Martyrs Asia

183

Libyan Jihad India Asia Southern Active USCFAFL 1 Movement Asia Libyan National India Asia Southern Active USCFAFL 1 Democratic Asia Movement Libyan National India Asia Southern Active USCFAFL 1 Grouping Asia Libyan National ATTF India Asia Southern Active USCFAFL 1 Salvation Asia Committee National Front for India Asia Southern Active USCFAFL 1 the Salvation of Asia politically Libya Macedonian India Asia Southern Active USCFAFL 1 Revolutionary Asia Organisation - Democratic Party for Macedonian National Unity Unikom (ethnic BLTF India Asia Southern Active USCFAFL 1 Albanians) Asia Azaouad Islamic- BSF India Asia Southern Active USCFAFL 1 Arab Front Asia Azaouad Popular India Asia Southern Active USCFAFL 1 Liberation Front Asia Azaouad Popular India Asia Southern Active USCFAFL 1 Movement Asia Azaouad India Asia Southern Active USCFAFL 1 Revolutionary Asia Army United Azaoud HUA India Asia Southern Active USCFAFL 1 Movements and Asia Fronts Justice Army of the India Asia Southern Active USCFAFL 1 Defenseless People Asia Popular JKLF India Asia Southern Active USCFAFL 1 Revolutionary Asia Army Zapatista National India Asia Southern Active USCFAFL 1 Liberation Asia Movement Popular Front India Asia Southern Active USCFAFL 1 Asia Republic of India Asia Southern Active USCFAFL 1 Transdniestr Asia Popular Front for MCC India Asia Southern Active USCFAFL 1 the Liberation of Asia Sakiet el Hamra and Rio de Oro Mozambican India Asia Southern Active USCFAFL 1 National Resistance Asia National Resistance NDF India Asia Southern Active USCFAFL 1 Movement B Asia Arakan Rohingya NLFT India Asia Southern Active USCFAFL 1 Islamic Front Asia

184

Kachin Democratic NSC India Asia Southern Active USCFAFL 1 Army N Asia Kachin PWG India Asia Southern Active USCFAFL 1 Independence Asia Army Karen Buddhist ULF India Asia Southern Active USCFAFL 1 Democracy Army A Asia Armenian Banglades Asia Southern Cease-fire USCFAFL 1 Liberation Army h Asia Jaish e IRP Tajikistan Asia Western Integrating USCFAFL 1 Mohammend Asia into governmen t Jeemah Islamiyah Tajikistan Asia Western Active USCFAFL 1 Asia Ansar Al-Islam Tajikistan Asia Western Active USCFAFL 1 Asia Lashkar I Jhangvi Tajikistan Asia Western Active USCFAFL 1 Asia Kurdish Kyrgyzsta Asia Western Active USCFAFL 1 Communist Party n Asia of Iran, Committee of the Revolutionary Toilers of Iranian Kordestan Nadeem Georgia Asia Western Active USCFAFL 1 Commando Asia Popular Front for Georgia Asia Western USCFAFL 1 Armed Resistance Asia Shi'ite Movement Georgia Asia Western Uncertain USCFAFL 1 of Pakistan Asia Bougainville Georgia Asia Western USCFAFL 1 Revolutionary Asia Army Moro Islamic Georgia Asia Western Cease-fire USCFAFL 1 Liberation Front Asia Moro National UNS Georgia Asia Western USCFAFL 1 Liberation Front O Asia National Georgia Asia Western USCFAFL 1 Democratic Front Asia Justice Commandos MLB Bahrain Asia Western Active USCFAFL 1 of the Armenian Asia Genocide People's BRA Papua Austral Australia Cease-fire USCFAFL 1 Combatants Group New ia in effect: Guinea 4/30/98 National Liberation BLA Austria Europe Central Active USCFAFL 1 Front of Kurdistan Europe People's Liberation VAP Austria Europe Central Active USCFAFL 1 Army of Kurdistan O Europe Harakat-ul-Ansar AIZ Germany Europe Central USCFAFL 1 Europe Jamaat ul-Fuqra Germany Europe Central USCFAFL 1 Europe

185

Muhajir Quami RAF Germany Europe Central Disbanded USCFAFL 1 Movement - Haqiqi Europe 4/21/98 Faction Muttahidda Quami RZ Germany Europe Central Dormant USCFAFL 1 Movement - Altaf Europe Faction United Popular Russia Europe Eastern Active USCFAFL 1 Action Movement- Europe Lautaro Tibetan Separatists Russia Europe Eastern USCFAFL 1 Europe Guyana National PF Moldova Europe Eastern USCFAFL 1 Service Europe Guyana People's Moldova Europe Eastern USCFAFL 1 Militia Europe All Tripura Tiger VMR Macedoni Europe Southeast USCFAFL 1 Force O- a, Europe DPM FYROM N Ananda Marg Macedoni Europe Southeast Active USCFAFL 1 a, Europe FYROM Shan United RO- Greece Europe Southeast Active USCFAFL, 4 Revolutionary 17 Europe EU, UK, US Army (Mong Tai Army) United Wa State Greece Europe Southeast Active USCFAFL 1 Army Europe Revolutionary Greece Europe Southeast Active USCFAFL 1 Armed Front Europe Azaouad Liberation Greece Europe Southeast Active USCFAFL 1 Front Europe Organisation de la MAS Greece Europe Southeast Active USCFAFL 1 Resistance Europe Revolutionary Greece Europe Southeast Active USCFAFL 1 Liberation Army of Europe North-Niger Saharan Patriotic Greece Europe Southeast Active USCFAFL 1 Liberation Front Europe Baluch People's ELA Greece Europe Southeast Active USCFAFL 1 Liberation Front Europe Baluch Students' Greece Europe Southeast Active USCFAFL 1 Organization - Europe Awami Paykar BR Italy Europe Southern Dormant USCFAFL, 2 Europe US Tudeh Italy Europe Southern Active USCFAFL 1 Europe Al-Dawa al- Italy Europe Southern Active USCFAFL 1 Islamiya Europe Communist Party Italy Europe Southern Active USCFAFL 1 militia Europe Cabinda Enclave ETA Spain Europe Southwest Active USCFAFL, 4 Liberation Front - Europe EU, UK, US Cabinda Armed

186

Forces Cabinda Enclave GRA Spain Europe Southwest Dormant USCFAFL, 3 Liberation Front - PO Europe EU, US Renovada National Union for Spain Europe Southwest Inactive USCFAFL 1 the Total Europe Independence of Angola Bavarian Liberation Spain Europe Southwest Uncertain USCFAFL 1 Army Europe Uighur Muslim Portugal Europe Southwest USCFAFL 1 Separatists Europe April 19 Movement Portugal Europe Southwest USCFAFL 1 Europe Peasant Self- FP-25 Portugal Europe Southwest USCFAFL 1 Defense Group of Europe Cordoba and Uraba Loyalist Volunteer CIRA United Europe Western Active USCFAFL, 3 Force Kingdom Europe EU, US Ulster Defense LVF United Europe Western Active USCFAFL, 3 Association Kingdom Europe EU, US Ulster Freedom UDA United Europe Western Cease-fire USCFAFL, 3 Fighters Kingdom Europe EU, US International Sikh IRA United Europe Western Active USCFAFL, 2 Youth Federation Kingdom Europe US Islamic Army of CAC United Europe Western Active USCFAFL 1 Aden Kingdom Europe Islamic Movement INLA United Europe Western Active USCFAFL 1 of Uzbekistan Kingdom Europe National Liberation UVF United Europe Western Cease-fire USCFAFL 1 Army - Colombia Kingdom Europe Revolutionary AD France Europe Western Dormant USCFAFL 1 Proletarian Army Europe Azorean Liberation FLN France Europe Western Active USCFAFL 1 Front C- Europe HW Azorean Nationalist FLN France Europe Western Active USCFAFL 1 Movement C-CH Europe Popular Forces of FLN France Europe Western Active, USCFAFL 1 the 25th of April KS Europe non- violent Islamic Tunisian CCC Belgium Europe Western Dormant USCFAFL 1 Front Europe Srpska Iraq Middle Middle AUS, US 2 Dobrovoljacka East East Garda (SDG) Srpska Garda Real IRA Tunisia Middle Middle Active USCFAFL 1 East East Red Hand Tunisia Middle Middle Active USCFAFL 1 Defenders East East Revolutionary Tunisia Middle Middle Active USCFAFL 1 Nuclei East East Asbat Al-Ansar FIT Tunisia Middle Middle Active USCFAFL 1 East East

187

Ansuman¨¦ Man¨¦ Polisa Morocco Middle Middle Active USCFAFL 1 rebellion rio East East Azar Khalistan FIGL Libya Middle Middle Active USCFAFL 1 Babbar Khalsa East East Force Bodo Liberation Libya Middle Middle Active USCFAFL 1 Tiger Force East East Bodo Security Libya Middle Middle Active USCFAFL 1 Force East East Dal Khalsa Libya Middle Middle Active USCFAFL 1 East East Dashmesh Libya Middle Middle Active USCFAFL 1 Regiment East East Garo National Libya Middle Middle USCFAFL 1 Front East East Harakat ul-Ansar Libya Middle Middle USCFAFL 1 East East Jamaat-e-Islam Libya Middle Middle USCFAFL 1 East East Jammu and Libya Middle Middle USCFAFL 1 Kashmir Liberation East East Front Hezbollah Gulf EIJ Egypt Middle Middle Active USCFAFL, 4 East East UK, AUS, US Islamic Jihad in Egypt Middle Middle USCFAFL 1 Hejaz East East Islamic Peninsula IG / Egypt Middle Middle Active USCFAFL 1 Movement for GAI East East Change - Jihad Wing Islamic Egypt Middle Middle USCFAFL 1 Revolutionary East East Organization Irish National GIA Algeria Middle Middle Active USCFAFL, 3 Liberation Army East East UK, US Red Hand AKA Algeria Middle Middle Uncertain USCFAFL 1 Commandos L East East Ulster Volunteer Algeria Middle Middle USCFAFL 1 Force East East Armed FIS/A Algeria Middle Middle Active USCFAFL 1 Commandos for IS East East National Liberation Palestine Islamic Yemen Middle Middle Active USCFAFL 1 Jihad - Shiqaqi East East faction United People's DHK Turkey Middle Middle Active USCFAFL, 5 Front of Nepal P/C, East East EU, UK, Dev JPN, US Sol Sipah-i-Sahaba PKK Turkey Middle Middle Active USCFAFL, 2 Pakistan East East EU Tupac Amaru ALA Turkey Middle Middle Dormant USCFAFL 1 Revolutionary East East Movement

188

Alex Boncayo ASA Turkey Middle Middle Dormant USCFAFL 1 Brigade LA East East New People's Army Turkey Middle Middle Active USCFAFL 1 East East Kurdistan Workers JCAG Turkey Middle Middle Dormant USCFAFL 1 Party East East Allied Democratic ERN Turkey Middle Middle Active USCFAFL 1 Forces K East East Lord's Resistance ARG Turkey Middle Middle Active USCFAFL 1 Army K East East Irish Republican TYK Turkey Middle Middle Active USCFAFL 1 Army B East East Eastern Turkistan Syria Middle Middle Active, but USCFAFL 1 Islamic Movement East East suppressed Harkat-i-Islami Sudan Middle Middle Active USCFAFL 1 East East Jamaat e Islami Sudan Middle Middle Active USCFAFL 1 East East National Islamic Sudan Middle Middle Active USCFAFL 1 Movement East East Northern Alliance NDA Sudan Middle Middle Active USCFAFL 1 East East Taliban Militia SSIM Sudan Middle Middle Active USCFAFL 1 East East United Islamic SAF Sudan Middle Middle Active USCFAFL 1 Front for the East East Salvation of Afghanistan Alliance for a Free SPLA Sudan Middle Middle Active USCFAFL 1 Kabylie East East Belmokhtar Group Sudan Middle Middle Active USCFAFL 1 East East Movement for Saudi Middle Middle Active USCFAFL 1 Democracy and Arabia East East Development National Saudi Middle Middle Active USCFAFL 1 Awakening Arabia East East Committee for Peace and Democracy National Council Saudi Middle Middle Active USCFAFL 1 for Rebuilding Arabia East East Chad National Council Saudi Middle Middle Active USCFAFL 1 for Recovery Arabia East East National Front for Saudi Middle Middle Active USCFAFL 1 the Renewal of Arabia East East Chad People's Saudi Middle Middle Active USCFAFL 1 Democratic Front Arabia East East Union of Saudi Middle Middle Active USCFAFL 1 Democratic Forces Arabia East East Lautaro Youth Saudi Middle Middle Active USCFAFL 1 Movement Arabia East East Manuel Rodriguez Saudi Middle Middle Active USCFAFL 1

189

Patriotic Front - Arabia East East Autonomous Manuel Rodriguez Saudi Middle Middle Active USCFAFL 1 Patriotic Front - Arabia East East Dissidents Red Sun SSP Pakistan Middle Middle Active USCFAFL, 2 East East US al-Jihad BPLF Pakistan Middle Middle Unclear USCFAFL 1 East East Islamic Group BSO- Pakistan Middle Middle Unclear USCFAFL 1 A East East Vanguards of Pakistan Middle Middle Active USCFAFL 1 Conquest East East Farabundo Marti Pakistan Middle Middle Active USCFAFL 1 National Liberation East East Front Movement for the MQM Pakistan Middle Middle Active USCFAFL 1 Autodetermination -H East East of the Island of Bioko Eritrean Liberation MQM Pakistan Middle Middle Active USCFAFL 1 Front East East Al-Ittihad al-Islami Pakistan Middle Middle Active USCFAFL 1 East East Action Directe PFAR Pakistan Middle Middle Unclear USCFAFL 1 East East Corsican National Pakistan Middle Middle Active USCFAFL 1 Liberation Front- East East Historic Wing Maoist Communist PKK Lebanon Middle Middle Dormant USCFAFL, 5 Center East East EU, UK, JPN, US Muslim Lebanon Middle Middle Active USCFAFL 1 Brotherhood East East National Lebanon Middle Middle Active USCFAFL 1 Democratic Front East East of Bodoland National Liberation Lebanon Middle Middle Active USCFAFL 1 Front of Tripura East East National Socialist Lebanon Middle Middle Active USCFAFL 1 Council of East East Nagaland People's War Lebanon Middle Middle Active USCFAFL 1 Group East East United Liberation Lebanon Middle Middle Active USCFAFL 1 Front of Assam East East Revolutionary Lebanon Middle Middle Active USCFAFL 1 Front for an East East Independent East Timor Gerakin Aceh Lebanon Middle Middle Active USCFAFL 1 Merdeka East East Organisasi Papua Lebanon Middle Middle USCFAFL 1 Merdek East East Al-Harakan al- MUI Lebanon Middle Middle Active USCFAFL 1

190

Islamiya East East Ansar-e Hezbollah Lebanon Middle Middle Active USCFAFL 1 East East Babak Khoramdin FAR Lebanon Middle Middle Dormant USCFAFL 1 Organisation L East East Banner of Kaveh Amal Lebanon Middle Middle Active USCFAFL 1 East East Democratic Party PFLP Lebanon Middle Middle Active USCFAFL 1 of Iranian East East Kurdistan Democratic Lebanon Middle Middle Active USCFAFL 1 Revolutionary East East Front for the Liberation of Arabistan Fadayan - Majority Lebanon Middle Middle Active USCFAFL 1 Faction East East Kurdish Jordan Middle Middle Active USCFAFL 1 Democratic Party East East of Iran Iranian Democratic ANO Israel Middle Middle Active USCFAFL, 5 Party of Kurdistan East East EU, UK, JPN, US Islamic Movement Israel Middle Middle Active USCFAFL, 5 of Kurdistan East East EU, UK, AUS, US Kurdistan Kach Israel Middle Middle Active USCFAFL, 4 Democratic Party East East EU, JPN, US Patriotic Union of PIJ Israel Middle Middle Active USCFAFL, 4 Kurdistan East East EU, UK, US Socialist Party PLF Israel Middle Middle Active, but USCFAFL, 4 militia East East restrained EU, JPN, US Supreme Council PFLP Israel Middle Middle Active USCFAFL, 4 for Islamic East East EU, JPN, Revolution US Supreme Council PFLP Israel Middle Middle Active USCFAFL, 4 for the Islamic -GC East East EU, JPN, Resistance in Iraq, US Badr Corps Turcoman Front DFLP Israel Middle Middle Active USCFAFL 1 Militia East East Democratic Front Israel Middle Middle Active USCFAFL 1 for the Liberation East East of Palestine Fatah Uprising Israel Middle Middle USCFAFL 1 East East Gush Emunim Israel Middle Middle Active USCFAFL 1 Underground East East Islamic Jihad Israel Middle Middle Active USCFAFL 1 East East May 15 OAA Israel Middle Middle Inactive USCFAFL 1 Organization S East East Organization of the PFLP Israel Middle Middle Active USCFAFL 1

191

Armed Arab -SC East East Struggle Popular Front for PSF Israel Middle Middle Dormant USCFAFL 1 the Liberation of East East Palestine-Special Command Popular Struggle TNT Israel Middle Middle USCFAFL 1 Front East East Terror Against Iraq Middle Middle Active USCFAFL 1 Terror East East Autonomists Iraq Middle Middle USCFAFL 1 East East Hammer Skinheads KDPI Iraq Middle Middle USCFAFL 1 Italia East East Third Position IMK Iraq Middle Middle USCFAFL 1 East East Blood Revenge KDP Iraq Middle Middle Active USCFAFL 1 Corps of the East East Partisan Volunteer Corps for the Independence of the Japanese Race Kakamaru-ha PUK Iraq Middle Middle Active USCFAFL 1 East East Middle Core Iraq Middle Middle USCFAFL 1 Faction, or Nucleus East East Sane Thinkers Iraq Middle Middle USCFAFL 1 School East East Jordanian Muslim SCIR Iraq Middle Middle Active USCFAFL 1 Brotherhood I East East Independence Iraq Middle Middle USCFAFL 1 East East Lao Liberation MEK/ Iran Middle Middle Active USCFAFL, 5 Army MKO East East EU, UK, , JPN, US PMOI , NLA Lao National Iran Middle Middle Uncertain USCFAFL 1 Liberation East East Movement United Lao Iran Middle Middle Active USCFAFL 1 National Liberation East East Front Al Ekhouwan al BKO Iran Middle Middle USCFAFL 1 Muslimin East East Al Gamaat Al Iran Middle Middle USCFAFL 1 Islamiyya East East Al Taqfeer Wal DPIK Iran Middle Middle Active USCFAFL 1 Hijra East East Army of Palestine Iran Middle Middle Uncertain USCFAFL 1 East East Fatah Iran Middle Middle USCFAFL 1 East East Hamas Iran Middle Middle USCFAFL 1

192

East East Hezbollah FMI Iran Middle Middle USCFAFL 1 East East Islamic Jehad Iran Middle Middle USCFAFL 1 East East Islamic Resistance KOM Iran Middle Middle Active USCFAFL 1 ALA East East Islamic Unification KDP Iran Middle Middle Active USCFAFL 1 Movement East East Jamaat Al Noor Iran Middle Middle Active USCFAFL 1 East East Lebanese Armed Iran Middle Middle Active USCFAFL 1 Revolutionary East East Faction Lebanese Iran Middle Middle Active USCFAFL 1 Resistance East East Detachments Popular Front for Iran Middle Middle Active USCFAFL 1 the Liberation of East East Palestine Usbat Al Ansar NLA Iran Middle Middle Active USCFAFL 1 East East Usbat Al Nour Iran Middle Middle USCFAFL 1 East East Iran Middle Middle USCFAFL 1 East East National Patriotic Iran Middle Middle USCFAFL 1 Front of Liberia East East Armenian Secret Bahrain Middle Middle Active USCFAFL 1 Army for the East East Liberation of Armenia Grey Wolves IFLB Bahrain Middle Middle Active USCFAFL 1 (Idealists) East East Armed Forces of Afghanist Middle Middle Active USCFAFL 1 National Liberation an East East Armed Forces of Afghanist Middle Middle Active USCFAFL 1 Popular Resistance an East East Army of God NIM Afghanist Middle Middle Active USCFAFL 1 an East East Aryan Nations Afghanist Middle Middle Active USCFAFL 1 an East East Guerrilla Forces of Afghanist Middle Middle Active USCFAFL 1 Liberation an East East Los Macheteros UIFS Afghanist Middle Middle Active USCFAFL 1 A an East East Hamas Yugoslavi Middle Middle Uncertain USCFAFL 1 a East East Kurdistan Workers U?K/ Turkey Middle Middle Active USCFAFL 1 Party KLA East East Revolutionary FAR Yugoslavi Middle Middle USCFAFL 1 People's Liberation K a East East Party/Front Egyptian Islamic LAK Egypt Middle Middle USCFAFL 1 Jihad Group East East

193

November 17 Yugoslavi Middle Middle USCFAFL 1 a East East Babbar Khalsa Yugoslavi Middle Middle USCFAFL 1 a East East Kahane Chai Yugoslavi Middle Middle USCFAFL 1 a East East Al-Aqsa Martyrs UFF Israel Middle Middle Cease-fire USCFAFL, 3 Brigade East East EU, US Salafist Group for Algeria Middle Middle Cease-fire USCFAFL 1 Call and Combat East East Militia Groups IG NA NA EU, UK, 4 JPN, US Mountaineer LET, NA NA EU, UK, 4 Militia LT AUS, US Organization of NA NA UK, AUS, 4 Volunteers for the US, JPN Puerto Rican Revolution People's NA NA EU, JPN, 3 Revolutionary US Commandos National Liberation ISYF NA NA EU, UK, 3 Movement JPN (Tupamaros) Red Flag IAA NA NA UK, AUS, 3 US United IMU NA NA UK, AUS, 3 Revolutionary US Front United Front for the GSPC NA NA UK, AUS, 3 Liberation of US Oppressed Races Popular Front for RIRA NA NA EU, US 2 the Liberation of Sakiet el Hamra and Rio de Oro Yemeni Tribesmen RHD NA NA EU, US 2 Beli Orlovi NA NA EU, US 2 Kosovo Liberation NA NA UK, AUS 2 Army Kosovo Republic NA NA UK, AUS 2 Armed Forces Liberation Army of JeM NA NA UK, US 2 Kosova National Movement JI NA NA UK, US 2 for the Liberation of Kosovo Srpski Cetnicki LJ NA NA AUS, US 2 Pokret Black Mamba ETIM NA NA CHN, US 2 Chimwenje NA NA USCFAFL 1 Al-Takfir and Al- NA NA Middle EU 1 Hijra East Great Islamic IBDA NA NA EU 1 Eastern Warriors -C

194

Front Holy Land NA NA EU 1 Foundation for Relief and Development Stichting Al Aqsa NA NA EU 1 Orange Volunteers OV NA NA EU 1 Revolutionary ELA NA NA EU 1 Popular Struggle Harakat HM NA NA UK 1 Mujahideen Harakat Ul- NA NA AUS 1 Mujahideen Jaish-i- NA NA AUS 1 Mohammend Jemaah Islamiyah NA NA AUS 1 The Eastern ETL NA NA CHN 1 Turkistan O Liberation Organization The World Uighur WUY NA NA CHN 1 Youth Congress C Communist Party CPP? NA NA US 1 of Philippines/New NPA People's Army Harakat ul- HUM NA NA US 1 Mujahidin Hizballah Middle NA NA US 1 East Al-Badhr NA NA US 1 Mujahedin Anti-Imperialist NTA NA NA US 1 Territorial Nuclei Army for the ALIR NA NA US 1 Liberation of Rwanda Cambodian CFF NA NA US 1 Freedom Fighters Harakat ul-Jihad-I- HUJI NA NA US 1 Islami Harakat ul-Jihad-I- HUJI- NA NA US 1 Islami/Bangladesh B Hizb-I Islami NA NA US 1 Gulbuddin Islamic NA NA US 1 International Peacekeeping Brigade Jamiat ul- NA NA US 1 Mujahedin Kumpulan KMM NA NA US 1 Mujahidin Malaysia Libyan Islamic NA NA US 1 Fighting Group

195

Moroccan Islamic GIC NA NA US 1 Combatant Group M New Red BR/P NA NA US 1 Brigade/Communis CC t Combatant Party People Against PAG NA NA US 1 Gangsterism and AD Drugs Riyadus-Salikhin NA NA US 1 Reconnaissance and Sabotage Battalion of Chechen Martyrs Special Purpose NA NA US 1 Islamic Regiment The Tunisian TCG NA NA US 1 Combatant Group Turkish Hizballah NA NA US 1 The East Turkistan ETIC NA NA CHN 1 Information Center Revolutionary NIPR NA NA US Proletarian Initiative Nuclei Basque Fatherland United North North Active; USCFAFL 1 and Liberty States Americ America low level a Gama'a al- FAL United North North Active; USCFAFL 1 Islamiyya N States Americ America low level a Lashkar e United North North Active; USCFAFL 1 Tayyaba/Pashan-e- States Americ America low level Ahle Hadis a Al Qaida United North North Active USCFAFL 1 States Americ America a Armed Islamic AN United North North Active USCFAFL 1 Group States Americ America a Revolutionary United North North Active; USCFAFL 1 Armed Forces of States Americ America low level Colombia a United Self- United North North Active; USCFAFL 1 Defense Forces of States Americ America low level Colombia a Aum Shinrikyo United North North Active USCFAFL 1 States Americ America a October First Anti- United North North Active USCFAFL 1 Fascist Resistance States Americ America Group a Liberation Tigers United North North Active; USCFAFL 1 of Tamil Eelam States Americ America low level a Continuity Irish United North North Active; USCFAFL 1 Republic Army States Americ America low level

196

a Anti-Imperialist SL Peru North North Active USCFAFL, 4 Patriotic Union Americ America EU, JPN, a US Alfaro Lives, MRT Peru North North Active USCFAFL, 2 Damnit! A Americ America US a Syrian Muslim FLQ Canada North North Inactive USCFAFL 1 Brotherhood Americ America a Tigers of the Gulf UPA Dominica South South USCFAFL 1 n Americ America Republic a Popular Front for BR Middle South South Uncertain USCFAFL 1 the Liberation of East Americ America Palestine-General a Command Shining Path Venezuel South South Uncertain USCFAFL 1 a Americ America a Abu Sayyaf Group MLN Uruguay South South Legal USCFAFL 1 Americ America political a party Horsemen FAR Nicaragua South South Active USCFAFL 1 Americ America a Cinchonero Popular Mexico South South Active USCFAFL 1 Liberation Americ America Movement a Comandos EPR Mexico South South Active USCFAFL 1 Operativos Americ America Especiales a Morazanist EZL Mexico South South Active, USCFAFL 1 Patriotic Front N Americ America making a peace Karen National MPL Honduras South South Unclear; USCFAFL 1 Union / Karen Americ America long quiet National Liberation a Army Karenni Army COE Honduras South South Active USCFAFL 1 S Americ America a Myanmar National FPM Honduras South South Inactive, USCFAFL 1 Democratic Americ America but intact Alliance Army a National FRP- Honduras South South Inactive USCFAFL 1 Democratic LZ Americ America Alliance Army a New Democratic Guyana South South Disbanded USCFAFL 1 Army Americ America a Rohingya Guyana South South Disbanded USCFAFL 1 Solidarity Americ America Organization a Shan State URN Guatemal South South Disarmed USCFAFL 1

197

Restoration Council G a Americ America in 1997 (Mong Tai Army) a Fighting Ansar of FML El South South Political USCFAFL 1 Allah N Salvador Americ America Party a Jamaat al-Adala al- AVC Ecuador South South Active USCFAFL 1 Alamiya Americ America (one a faction) Legion of the GCP Ecuador South South Formed in USCFAFL 1 Martyr Abdullah Americ America 11/1998 al-Huzaifi a Movement for Ecuador South South Officially USCFAFL 1 Islamic Change Americ America disbanded a Armed Forces M-19 Cuba South South Inactive USCFAFL 1 Revolutionary Americ America Council a Revolutionary Cuba South South Inactive USCFAFL 1 United Front Americ America a Rahanwein FAR Colombia South South Active USCFAFL, 3 Resistance Army C Americ America EU, US a Somali Democratic AUC Colombia South South Active USCFAFL, 3 Alliance Americ America EU, US a Somali Democratic ELN Colombia South South Active USCFAFL, 2 Association Americ America US a Somali National M-19 Colombia South South A few are USCFAFL 1 Alliance Americ America active a Somali National ACC Colombia South South Active USCFAFL 1 Front U Americ America a Somali National EPL Colombia South South Active USCFAFL 1 Movement Americ America a Somali Patriotic FRF Colombia South South Active (as USCFAFL 1 Movement Americ America bandits) a United Somali MJL Chile South South Dormant USCFAFL 1 Congress Americ America a United Somali FPM Chile South South USCFAFL 1 Front R/A Americ America a Afrikaaner FPM Chile South South Active USCFAFL 1 Weestand R/D Americ America (Chile's Beweeging & Boer a only) Attack Troops Iraultza MIR Chile South South Dormant USCFAFL 1 Americ America a Those of the North MAP Chile South South USCFAFL 1

198

U-L Americ America a Islamic Liberation ELN Bolivia South South Dormant, USCFAFL 1 Party Americ America 1996 a Islamic Tendency EGT Bolivia South South Possibly USCFAFL 1 Party K Americ America active a Sources: USCFAFL = United States Committee For A Free Lebanon; US = United State Dept. Report; UK = Uinited Kingdom Government; EU = European Union; AUS = Australia Government; CHN = PRC Government; JPN = Japan Government

URL Group Name Location http://1osamabinladen.5u.com/ Al-Qaeda Unknown http://daawaparty.com/ Islamic Dawa Party Iraq http://impact.users.netlink.co.uk/namir/ National Movement of Iranian Resistance Iran namirm.html http://jorgevinhedo.sites.uol.com.br/in Markaz Ad-Dawa Wal Irshad Pakistan dex.html http://www.abrarway.com/ Islamic Jihad of Palestine Palestine http://www.alfida.jeeran.com/ Al-Qaeda Unkown http://www.alokab.com/ Salafi Group Unknown http://www.alsakifah.org/ Salafi Group Unknown http://www.ansar.ws Al-Ansar Unknown http://www.cihad.net/ Salafi Unknown http://www.clearguidance.com/ Salafi group Unkown http://www.dhkc.info/ Revolutionary Peoples Liberation Front Turkey http://www.dhkc.net/ Revolutionary Peoples Liberation Front Turkey http://www.etehadefedaian.org/ The Union Of People's Fedaian Of Iran Iran http://www.expliciet.nl/ Hizb-ut Tahrir Unknown http://www.ezzedeen.net/ Hamas Palestine http://www.fadai.org/ The Organization of Iranian People's Iran Fedaian (Majority) http://www.fadaian.org/ The Organization of Iranian People's Iran Fedaian (Majority) http://www.h-alali.net/ Salafi Group Unknown http://www.hilafet.com/ Hizb-ut Tahrir Unknown http://www.hizbollah.tv/ Hizballah Lebanon http://www.hizb-ut-tahrir.org/ Hizb-ut Tahrir Unknown http://www.infopalestina.com/ Hamas Palestine http://www.intiqad.com/ Hizballah Lebanon http://www.iran.mojahedin.org/ People's Mojahedin of Iran Iran http://www.iran-e-azad.org/ National Council of Resistance of Iran Iran http://www.iranncrfac.org/ National Council of Resistance of Iran Iran http://www.islamicdawaparty.org/ Islamic Dawa Party Iraq http://www.israel-wat.com/ Unknown Unknown http://www.jihadunspun.com/ Salafi Group Unknown http://www.kahane.org/ Kahane Chai Israel http://www.kavkazcenter.net/ Chechen Rebels Russia http://www.kdp.info/ Kurdistan Democratic Party-Iraq Iraq http://www.kdp.pp.se/ Kurdistan Democratic Party Iraq http://www.kdp.pp.se/ Kurdistan Democratic Party-Iraq Iraq

199

http://www.khilafah.com/ Hizb-ut Tahrir Unknown http://www.kurdistanmedia.com/ Democratic Party of Iranian Kurdistan Iran http://www.maktab-al-jihad.com/ Salafi Group Unknown http://www.manartv.org/ Hizballah Lebanon http://www.moqawama.tv/ Hizballah Lebanon http://www.nasrollah.org/ Hizballah Lebanon http://www.nehzateazadi.org/ Freedom Movement of Iran Iran http://www.pdk-iran.org/ Democratic Party of Iranian Kurdistan Iran http://www.puk.org/ Patriotic Union Of Kurdistan Iraq http://www.qoqaz.com/ Chechen Rebels Russia http://www.qudsway.com/ Islamic Jihad Palestine http://www.rantisi.net/ Hamas Palestine http://www.sahwah.com/ Salafi Group Unknown http://www.sciri.btinternet.co.uk/ The Supreme Council for the Islamic Iraq Revolution in Iraq http://www.shareeah.org/ Al-Qaeda Unkown http://www.siahkal.com/ The Iranian People's Fadaee Guerrillas Iran http://www.specialforce.net/ Hizballah Lebanon http://www.tanzeem.org/ Tanzeem Islami Pakistan http://www.tawhed.ws/ Al-Qaeda Unknown http://www.tudehpartyiran.org/ Tudeh Party Iran http://www.ummahnews.com/ Salafi Group Unknown http://www.wpiran.org/ Worker-Communist Party of Iran Iran http://www.wsahara.net/ Polisario Morocco http://download.specialforce.net Hizballah Lebanon http://palestine-info-urdu.com Hamas Palestine http://web.manartv.org Hizballah Lebanon http://www.abrarway.com Islamic Jihad Palestine http://www.al-fateh.net Hamas Palestine http://www.alokab.com Salafi Group Unknown http://www.alsakifah.org Salafi Group Unknown http://www.cihad.net Salafi Unknown http://www.clearguidance.com Salafi group Unkown http://www.expliciet.nl Hizb-ut Tahrir Unknown http://www.ezzedeen.net Hamas Palestine http://www.h-alali.net Salafi Group Unknown http://www.hilafet.com Hizb-ut Tahrir Unknown http://www.hizbollah.tv Hizballah Lebanon http://www.hizb-ut-tahrir.org Hizb-ut Tahrir Unknown http://www.infopalestina.com Hamas Palestine http://www.intiqad.com Hizballah Lebanon http://www.jihadunspun.com Salafi Group Unknown http://www.kataebalaqsa.com Al-Aqsa Martyrs Brigade Plalestine http://www.kavkazcenter.net Chechen Rebels Russia http://www.khilafah.com Hizb-ut Tahrir Unknown http://www.maktab-al-jihad.com Salafi Group Unknown http://www.moqawama.tv Hizballah Lebanon http://www.nasrollah.org Hizballah Lebanon http://www.qoqaz.com Chechen Rebels Russia http://www.qudsway.com Islamic Jihad Palestine http://www.rantisi.net Hamas Palestine http://www.sahwah.com Salafi Group Unknown http://www.shareeah.org Al-Qaeda Unkown http://www.tanzeem.org Tanzeem Islami Pakistan

200

http://www.ummahnews.com Salafi Group Unknown http://arabhackerz.8m.com/ Unknown Unknown http://wps.jeeran.com Hamas Palestine http://www.muhajiroun.com/ Salafi Group Unknown http://www.almaqreze.com/ Salafi Group Saudi Arabia http://www.islammemo.cc/ Salafi Group Unknown http://www.ansar-sonnah.8m.com/ Salafi Group Iraq http://www.geocities.com/salafiahweb/ Salafi Group For Call and Combat Algeria http://www.muslman.com/ Salafi Group Unknown http://www.freewebs.com/abuomar/ Salafi Group Unknown http://www.qal3ati.net/vb/ Salafi Group Unknown http://www.ajnad.50megs.com/ Salafi Group Iraq http://www.qawem.org Hamas Palestine http://www.kataebq.com Qassam Brigade-Military wing of Hamas Palestine http://www.sabiroon.org Hamas Palestine http://www.palestine-info.info Hamas Palestine http://www.gamla.org.il Supporters of Jewish Extremists Israel http://audio.kavkazcenter.com Chechen Rebels Russia http://old.kavkazcenter.com Chechen Rebels Russia http://www.kavkaz.tv Chechen Rebels Russia http://www.kavkaz.uk.com Chechen Rebels Russia http://www.kavkaz.org.uk Chechen Rebels Russia http://www.kavkazcenter.com Chechen Rebels Russia http://www.kavkazcenter.info Chechen Rebels Russia http://www.chechen.org Chechen Rebels Russia http://www.chechnya.biz Chechen Rebels Russia http://ichkeria.dk Chechen Rebels Russia http://serlo.prv.pl Chechen Rebels Russia http://www.chechnya.250x.com Chechen Rebels Russia http://www.ichkeria.250x.com Chechen Rebels Russia http://www.takbir.by.ru Chechen Rebels Russia http://vaynahchat.com Chechen Rebels Russia http://www.shamilonline.tk Chechen Rebels Russia http://www.kavkazchat.com Chechen Rebels Russia http://www.zurho.tk Chechen Rebels Russia http://www.zamir-j.de Chechen Rebels Russia http://www.al-nahda.org Chechen Rebels Russia http://www.ummah.ru Chechen Rebels Russia http://al-halifat.narod.ru Chechen Rebels Russia http://jihadfilislam.by.ru Chechen Rebels Russia http://www.salam-alejkum.narod.ru Chechen Rebels Russia http://amir-mumin.boom.ru Chechen Rebels Russia http://ahlbayt.by.ru Chechen Rebels Russia http://www.bedir.ru Chechen Rebels Russia http://www.tawhid.narod.ru Chechen Rebels Russia http://www.daymohk.info Chechen Rebels Russia http://manhaj.narod.ru Chechen Rebels Russia http://www.peshera.pochta.ru Chechen Rebels Russia http://www.ramzan.kazan.ws Chechen Rebels Russia http://nsudenko.narod.ru Chechen Rebels Russia http://www.engenoi.kazan.ws Chechen Rebels Russia http://www.al-halifat.narod.ru Chechen Rebels Russia http://www.jiubila.com Chechen Rebels Russia

201

http://www.zabarsh.nm.ru Chechen Rebels Russia http://chechenstar.narod.ru Chechen Rebels Russia http://www.resistance.by.ru Chechen Rebels Russia http://www.ichkeriya2004.narod.ru Chechen Rebels Russia http://www.gazavat.narod.ru Chechen Rebels Russia http://www.bachi-yurt.tk Chechen Rebels Russia http://almatinec.boxmail.biz Chechen Rebels Russia http://pyti-yti.narod.ru Chechen Rebels Russia http://www.daimokh.narod.ru Chechen Rebels Russia http://www.vaynah.com Chechen Rebels Russia

202

APPENDIX C

US DOMESTIC EXTREMIST FORUMS

Category Group Forum Name Forum URL Founder Found Date Name Neo-Nazis National American Yahoo http://groups.yahoo.com/group/amer americannationalsocialistgroup- Nov 26, 2003 Alliance National icannationalsocialistgroup/ [email protected] Socialist Group Neo-Nazis National Klandestine MSN http://groups.msn.com/Klandestine RButler64 Alliance Knights Knights/home.htm Neo-Nazis National Resistencia MSN http://groups.msn.com/ResistenciaA Alliance Aria ria/home.htm Neo-Nazis National NSM World Yahoo http://groups.yahoo.com/group/nsm nsmworld- Jan 12, 2004 Socialist world/ [email protected] Movement Neo-Nazis National talk.politics.th Google http://groups.google.com.mx/groups Socialist eory ?hl=es&lr=&group=talk.politics.the Movement ory Neo-Nazis National alt.revisionism Google http://groups.google.com.mx/groups Socialist ?hl=es&lr=&group=alt.revisionism Movement Neo-Nazis National alt.politics.whi Google http://groups.google.com/groups?hl Socialist te-power =en&lr=&c2coff=1&group=alt.polit Movement ics.white-power Neo-Nazis National alt.politics.nati Google http://groups.google.com/groups?hl Socialist onalism.white =en&lr=&c2coff=1&group=alt.polit Movement ics.nationalism.white Neo-Nazis National alt.conspiracy Google http://groups.google.com.mx/groups Socialist ?hl=es&lr=&group=alt.conspiracy Movement Neo-Nazis National National MSN http://groups.msn.com/NationalSoci

203

Socialist Socialist alistMovement/home.htm Movement Movement Neo-Nazis N/A Neo-Nazi Yahoo http://groups.yahoo.com/group/Neo- Neo-Nazi- Jan 14, 2004 Nazi/ [email protected] Neo-Nazis N/A Angelic_Adolf Yahoo http://groups.yahoo.com/group/Ang Angelic_Adolf- Dec 10, 2002 elic_Adolf/ [email protected] Neo-Nazis N/A africanh8ters Yahoo http://groups.yahoo.com/group/afric africanh8ters88- Jan 7, 2003 anh8ters88/ [email protected] Neo-Nazis N/A naziwoman Yahoo http://groups.yahoo.com/group/nazi naziwoman- Dec 14, 2001 woman/ [email protected] Neo-Nazis N/A smashnazism Yahoo http://groups.yahoo.com/group/sma smashnazism- Jan 26, 2002 shnazism/messages/81 [email protected] Neo-Nazis N/A thejapanesenaz Yahoo http://groups.yahoo.com/group/theja thejapanesenazis- May 12, 2003 is panesenazis/ [email protected] Neo-Nazis N/A ilovewhitefolk Yahoo http://groups.yahoo.com/group/ilov Jul 23, 2002 s3 ewhitefolks3/ Neo-Nazis N/A INVISIBLE_E Yahoo http://groups.yahoo.com/group/INV Jan 18, 2002 MPIRE ISIBLE_EMPIRE Neo-Nazis N/A misc.activism. Google http://groups.google.com/groups?hl militia =en&lr=&c2coff=1&group=misc.ac tivism.militia Neo-Nazis N/A alt.politics.nati Google http://groups.google.com/groups?hl onalism.white =en&lr=&c2coff=1&group=alt.polit ics.nationalism.white Neo-Nazis N/A misc.activism. Google http://groups.google.com/groups?hl progressive =en&lr=&c2coff=1&group=misc.ac tivism.progressive Neo-Nazis N/A sci.skeptic Google http://groups.google.com/groups?hl =en&lr=&c2coff=1&group=sci.ske ptic Neo-Nazis N/A alt.politics.whi Google http://groups.google.com/groups?hl te-power =en&lr=&c2coff=1&group=alt.polit ics.white-power Neo-Nazis N/A alt.skinheads Google http://groups.google.com/groups?hl =en&lr=&c2coff=1&group=alt.skin heads

204

Neo-Nazis N/A nazis of 2004 AOL http://groups.aol.com/sknhdsn?mmc KORNvsSLIPKNOT1 h_=0 Neo-Nazis N/A Aryan Raiders MSN http://groups.msn.com/AryanRaider ARFounder Feb. 11 2004 s/ Neo-Nazis N/A NEONAZI MSN http://groups.msn.com/NEONAZI/ X Seether X Neo-Nazis Women for Women for US http://www.stormfront.org/ Aryan Unity Aryan Unity WEB Neo-Nazis American American Nazi US http://www.nazi.org/current/forum/ Nazi Party Party WEB Neo-Nazis American American Nazi US http://www.nationalist.org/forum/in Nazi Party Party WEB dex.php Neo-Nazis American American Nazi US http://www.whiterevolution.com/for Nazi Party Party WEB um14/ White World World_Knight Yahoo http://groups.yahoo.com/group/Wor World_Knights- Nov 6, 2002 Supremacist Knights of s ld_Knights/ [email protected] the Ku Klux Klan White World aryannationskn Yahoo http://groups.yahoo.com/group/arya aryannationsknights- Apr 14, 2004 Supremacist Knights of ights nnationsknights/messages/238 [email protected] the Ku Klux Klan White World UK- Yahoo http://groups.yahoo.com/group/UK- UK-WHITEKNIGHTS- Jul 23, 2004 Supremacist Knights of WhiteKnights WHITEKNIGHTS/ [email protected] the Ku Klux Klan White World misc.activism. Google http://groups.google.com/groups?hl Supremacist Knights of progressive =en&lr=&c2coff=1&group=misc.ac the Ku Klux tivism.progressive Klan White World alt.skinheads Google http://groups.google.com/groups?hl Supremacist Knights of =en&lr=&c2coff=1&group=alt.skin the Ku Klux heads Klan White World alt.politics.nati Google http://groups.google.com/groups?hl Supremacist Knights of onalism.white =en&lr=&c2coff=1&group=alt.polit the Ku Klux ics.nationalism.white

205

Klan White World alt.flames.nigg Google http://groups.google.com/groups?hl Supremacist Knights of ers =en&lr=&c2coff=1&group=alt.flam the Ku Klux e.niggers Klan White World alt.politics.whi Google http://groups.google.com/groups?hl Supremacist Knights of te-power =en&lr=&c2coff=1&group=alt.polit the Ku Klux ics.white-power Klan White World KKK AOL http://groups.aol.com/klklxkln4?m Birdhouseshaun18 Supremacist Knights of mch_=0 the Ku Klux Klan White National National MSN http://groups.msn.com/NationalOffi Supremacist Office of Office of The ceofTheWorldKnightsoftheKKK/ho The World World Knights me.htm Knights of of the KKK the KKK White Bayou Bayou Knights US http://www.bayouknights.org/forum Supremacist Knights of of the Ku Klux WEB the Ku Klux Klan Klan White Imperial Imperial Klans US http://www.whiteriderrecords.com/ Supremacist Klans of of America WEB Community/ America White North North Georgia US http://www.theklan.com/forum/inde Supremacist Georgia White Knights WEB x.php White of the Ku Klux Knights of Klan the Ku Klux Klan White Southern Southern US http://www.swkkkk.org/members/in Supremacist White White Knights WEB dex.html Knights of of the Ku Klux the Ku Klux Klan Klan White Earth Earth US http://www.animalliberationfront.co

206

Supremacist Liberation Liberation WEB m/phpBB2b/index.php Front Front White Keystone misc.activism. Google http://groups.google.com/groups?hl Supremacist State progressive =en&lr=&c2coff=1&group=misc.ac Skinheads tivism.progressive White Keystone alt.skinheads Google http://groups.google.com/groups?hl Supremacist State =en&lr=&c2coff=1&group=alt.skin Skinheads heads White Greater Greater Things Yahoo http://groups.yahoo.com/group/Grea GreaterThingsMinistries- Sep 30, 2002 Supremacist Ministries Ministries terThingsMinistries/ [email protected] International International White New Black New Black Yahoo http://groups.yahoo.com/group/NBP NBPP-National- Jun 18, 2003 Supremacist Panther Panther Party P-National/ [email protected] Party for for Self- Self-Defense Defense White The White texaskkk Yahoo http://groups.yahoo.com/group/alfan texaskkk- Mar 1, 2004 Supremacist Knights of imalliberationfront/ [email protected] Texas White The White texasklan Yahoo http://groups.yahoo.com/group/texa texasklan- May 19, 2004 Supremacist Knights of sklan/ [email protected] Texas White The White us.legal Google http://groups.google.com/groups?hl Supremacist Knights of =en&lr=&c2coff=1&group=us.legal Texas White The White misc.activism. Google http://groups.google.com/groups?hl Supremacist Knights of progressive =en&lr=&c2coff=1&group=misc.ac Texas tivism.progressive White The White alt.flame.nigge Google http://groups.google.com/groups?hl Supremacist Knights of rs =en&lr=&c2coff=1&group=alt.flam Texas e.niggers White The White alt.politics.nati Google http://groups.google.com/groups?hl Supremacist Knights of onalism.white =en&lr=&c2coff=1&group=alt.polit Texas ics.nationalism.white White The White alt.politics.whi Google http://groups.google.com/groups?hl Supremacist Knights of te-power =en&lr=&c2coff=1&group=alt.polit Texas ics.white-power

207

White CNKKKK CNKKKK Yahoo http://groups.yahoo.com/group/CN CNKKKK- Dec 12, 2003 Supremacist KKKK/ [email protected] White Southern michiganklanb Yahoo http://groups.yahoo.com/group/mic michiganklanboyz- Sep 20, 2001 Supremacist White oyz higanklanboyz/ [email protected] Knights Florida White Southern talks.politics.m Google http://groups.google.com/groups?hl Supremacist White isc =en&lr=&c2coff=1&group=talk.pol Knights itics.misc Florida White Southern misc.activism. Google http://groups.google.com/groups?hl Supremacist White progressive =en&lr=&c2coff=1&group=misc.ac Knights tivism.progressive Florida White Texas League_Of_Th Yahoo http://groups.yahoo.com/group/LOS Oct 19, 2003 Supremacist League of e_South_Texa _Texas/ the South s White Texas alt.thought.sou Google http://groups.google.com.mx/groups Supremacist League of thern ?hl=es&lr=&group=alt.thought.sout the South hern Black Separatist United NUWAUBU Yahoo http://groups.yahoo.com/group/nuw nuwauburightknowledge- Feb 1, 1999 Nuwaubian RIGHT auburightknowledge/ [email protected] Nation of KNOWLEDG Moors E Black Separatist United soc.culture.afri Google http://groups.google.com.mx/groups Nuwaubian can.american, ?hl=es&lr=&group=soc.culture.afri Nation of can.american Moors Black Separatist United alt.politics.nati Google http://groups.google.com.mx/groups Nuwaubian onalism.black? ?hl=es&lr=&group=alt.politics.nati Nation of onalism.black Moors Black Separatist United alt.flame.nigge Google http://groups.google.com/groups?hl Nuwaubian rs =en&lr=&c2coff=1&group=alt.flam Nation of e.niggers Moors Black Separatist United alt.niggers Google http://groups.google.com.mx/groups

208

Nuwaubian ?hl=es&lr=&group=alt.niggers Nation of Moors Black Separatist New Black New Black Yahoo http://groups.yahoo.com/group/new newblackpantherparty- Mar 19, 2001 Panther Panther Party blackpantherparty/ [email protected] Party Black Separatist New Black alt.politics Google http://groups.google.com.mx/groups Panther ?hl=es&lr=&group=alt.politics Party Black Separatist New Black misc.activism. Google http://groups.google.com/groups?hl Panther progressive? =en&lr=&c2coff=1&group=misc.ac Party tivism.progressive Black Separatist New Black soc.culture.usa Google http://groups.google.com.mx/groups Panther ?hl=es&lr=&group=soc.culture.usa Party Black Separatist New Black alt.activism.* Google http://groups.google.com.mx/groups Panther ?hl=es&lr=&group=alt.activism Party Black Separatist New Black soc.culture.cub Google http://groups.google.com.mx/groups Panther a ?hl=es&lr=&group=soc.culture.cub Party a Black Separatist New Black Arizona Black AOL http://groups.aol.com/blackprotect? Shadrachn Panther Panther Party mmch_=0 Party Black Separatist New Black New Black MSN http://groups.msn.com/Countercolo Panther Panther Party nistAlliance/home.htm Party Black Separatist New Black misc.activism. Google http://groups.google.com/groups?hl Panther progressive =en&lr=&c2coff=1&group=misc.ac Party for tivism.progressive Self-Defense Christian Identity Church of ARYAN Yahoo http://groups.yahoo.com/group/arya aryannationsknights- Apr 14, 2004 the Sons of NATIONS nnationsknights/ [email protected] Yhvh KNIGHTS Christian Identity Church of alt.flame.fucki Google http://groups.google.com.mx/groups the Sons of ng.faggots ?hl=es&lr=&group=alt.flame.fuckin Yhvh g.faggots

209

Christian Identity Church of alt.flame.faggo Google http://groups.google.com.mx/groups the Sons of ts ?hl=es&lr=&group=alt.flame.faggot Yhvh s Christian Identity Westboro Peace Love Yahoo http://groups.yahoo.com/group/call callmefred- Dec 19, 2001 Baptist And Unity mefred/ [email protected] Church Topeka Militia California California Yahoo http://groups.yahoo.com/group/Cali CaliforniaMilitia- Jul 8, 2004 Militia Militia forniaMilitia [email protected] Militia California talk.politics.gu Google http://groups.google.com.mx/groups Militia ns ?hl=es&lr=&group=talk.politics.gu ns Militia California soc.culture.usa Google http://groups.google.com.mx/groups Militia ?hl=es&lr=&group=soc.culture.usa Militia California California US http://www.geocities.com/CapitolHi Militia Militia WEB ll/Congress/2608/forum.html Militia Michigan mmcwolverine Yahoo http://groups.yahoo.com/group/mm mmcwolverines- Feb 17, 2000 Militia s ¡¤ 10th cwolverines/ [email protected] Home Page Brigade Militia Michigan misc.activism. Google http://groups.google.com.mx/groups Militia militia ?hl=es&lr=&group=misc.activism. Home Page militia Militia Pennsylvani 7th Aerial Yahoo http://groups.yahoo.com/group/7th 7thAOSquadron- Dec 7, 2002 a State Obsever AOSquadron/ [email protected] Military Squadron Reserve Militia Sons of Southern Sons Yahoo http://groups.yahoo.com/group/ssol/ [email protected] May 11, 1999 Liberty Of Liberty Militia Militia Sons of misc.activism. Google http://groups.google.com.mx/groups Liberty militia ?hl=es&lr=&group=misc.activism. Militia militia Militia Sons of Sons of AOL http://groups.aol.com/libertyuson?m PlayaSoccer12 Liberty Liberty mch_=0 Militia Militia Sons of The Sons of AOL http://groups.aol.com/libertygang99 LONEWOLFXT89 Liberty Liberty 999?mmch_=0

210

Militia Militia Sons of ?-=(SoL)=- MSN http://groups.msn.com/SoLClan -=(SoL)=-Iceman Liberty Clan Militia Militia Sons of Pennsylvania MSN http://groups.msn.com/Pennsylvania Cowboy4Him Sept. 8 , 2004 Liberty Committee Of CommitteeOfSafety/ Militia Safety Neo-Confederate South South Carolina Yahoo http://groups.yahoo.com/group/sclo [email protected] Mar 15, 2003 Carolina League of the s/ League of South the South Neo-Confederate South alt.flame.nigge Google http://groups.google.com/groups?hl Carolina rs =en&lr=&c2coff=1&group=alt.flam League of e.niggers the South Neo-Confederate Council of Citizens Yahoo http://groups.yahoo.com/group/ccnu [email protected] Mar 17, 1999 Conservativ Councils News / e Citizens Update Neo-Confederate Council of alt.flame.nigge Google http://groups.google.com/groups?hl Conservativ rs =en&lr=&c2coff=1&group=alt.flam e Citizens e.niggers Neo-Confederate Council of alt.politics.de Google http://groups.google.com.mx/groups Conservativ mocrats ?hl=es&lr=&group=alt.politics.dem e Citizens ocrats Others Animal Animals have MSN http://groups.msn.com/AnimalsHav Liberation rights too eRightsToo/home.htm Front Others Animal misc.activism. Google http://groups.google.com/groups?hl Liberation progressive =en&lr=&c2coff=1&group=misc.ac Front tivism.progressive Others Animal Animal Yahoo http://groups.yahoo.com/group/alfan alfanimalliberationfront- Mar 27, 2000 Liberation Liberation imalliberationfront/ [email protected] Front Front Others Animal aus.politics Google http://groups.google.com.mx/groups Liberation ?hl=es&lr=&group=aus.politics Front

211

Others American American US http://www.fascistforum.com/ Fascist Fascist WEB Movement Movement

212

APPENDIX D

PART OF SOURCE CODES OF EDONKEY SPIDER AGENTS

DownloadClient.cpp void CUpDownClient::SetDownloadState(EDownloadState nNewState, LPCTSTR pszReason){ if (m_nDownloadState != nNewState){ switch ( nNewState ) { case DS_CONNECTING: m_dwLastTriedToConnect = ::GetTickCount(); break ; case DS_TOOMANYCONNSKAD: //This client had already been set to DS_CONNECTING. //So we reset this time so it isn't stuck at TOOMANYCONNS for 20mins. m_dwLastTriedToConnect = ::GetTickCount()-20*60*1000; break ; case DS_WAITCALLBACKKAD: case DS_WAITCALLBACK: break ; case DS_NONEEDEDPARTS: // Since tcp asks never sets reask time if the result is DS_NONEEDEDPARTS // If we set this, we will not reask for that file until some time has passed. SetLastAskedTime(); //DontSwapTo(reqfile); default : switch ( m_nDownloadState ) { case DS_WAITCALLBACK: case DS_WAITCALLBACKKAD: break ; default : m_dwLastTriedToConnect = ::GetTickCount()- 20*60*1000; break ; } break ; }

if ( nNewState==DS_LOWTOLOWIP && m_bSupportNatTraverse ) //huby edit { m_fileReaskTimes.SetAt( reqfile, ::GetTickCount() + 5*60*1000 ); m_iErrTimes++; m_iErrTimes++; }

if (reqfile){ if (nNewState == DS_DOWNLOADING){

213

if (thePrefs.GetLogUlDlEvents()) AddDebugLogLine(DLP_VERYLOW, false , _T( "Download session started. User: %s in SetDownloadState(). New State: %i" ), DbgGetClientInfo(), nNewState); reqfile->AddDownloadingSource( this ); } else if (m_nDownloadState == DS_DOWNLOADING){ reqfile->RemoveDownloadingSource( this ); } }

if (nNewState == DS_DOWNLOADING && socket){ socket->SetTimeOut(CONNECTION_TIMEOUT*4); }

if (m_nDownloadState == DS_DOWNLOADING ){ if (socket) socket->SetTimeOut(CONNECTION_TIMEOUT);

if (thePrefs.GetLogUlDlEvents()) { switch ( nNewState ) { case DS_NONEEDEDPARTS: pszReason = _T( "NNP. You don't need any parts from this client." ); }

if (thePrefs.GetLogUlDlEvents()) AddDebugLogLine(DLP_VERYLOW, false , _T( "Download session ended: %s User: %s in SetDownloadState(). New State: %i, Length: %s, Payload: %s, Transferred: %s, Req blocks not yet completed: %i." ), pszReason, DbgGetClientInfo(), nNewState, CastSecondsToHM(GetDownTimeDifference( false )/1000), CastItoXBytes(GetSessionPayloadDown(), false , false ), CastItoXBytes(GetSessionDown(), false , false ), m_PendingBlocks_list.GetCount()); }

ResetSessionDown();

// -khaos--+++> Extended Statistics (Successful/Failed Download Sessions) if ( m_bTransferredDownMini && nNewState != DS_ERROR ) thePrefs.Add2DownSuccessfulSessions(); // Increment our counters for successful sessions (Cumulative AND Session) else thePrefs.Add2DownFailedSessions(); // Increment our counters failed sessions (Cumulative AND Session) thePrefs.Add2DownSAvgTime(GetDownTimeDifference()/1000); // <-----khaos-

m_nDownloadState = (_EDownloadState)nNewState;

ClearDownloadBlockRequests();

m_nDownDatarate = 0; m_AvarageDDR_list.RemoveAll(); m_nSumForAvgDownDataRate = 0;

if (nNewState == DS_NONE){ delete [] m_abyPartStatus;

214

m_abyPartStatus = NULL; m_nPartCount = 0; } if (socket && nNewState != DS_ERROR ) socket->DisableDownloadLimit(); } m_nDownloadState = (_EDownloadState)nNewState; if ( GetDownloadState() == DS_DOWNLOADING ){ if ( IsEmuleClient() ) SetRemoteQueueFull( false ); SetRemoteQueueRank(0); SetAskedCountDown(0); } UpdateDisplayedInfo( true );

//01/19/2008, Guanpi Lai CStdioFile* SourceFile = new CStdioFile(); switch (nNewState) {

case DS_DOWNLOADING: case DS_ONQUEUE: case DS_CONNECTED: case DS_NONEEDEDPARTS: case DS_REMOTEQUEUEFULL: if (m_pszUsername && m_achUserHash && reqfile) { MYSQL *hnd = thePrefs.GetMysqlConnection(); if (hnd!=NULL) { CString username(m_pszUsername); username.Replace(_T( "'" ),_T( "\\'" )); CString record; record.Format(_T( "insert ignore into %s set hash='%s',nick='%s',user_ip_port='%s',user_hash='%s',server_ip_port='%s',type = 0, added=now()" ),thePrefs.GetMysqlTable(), md4str(reqfile->GetFileHash()), username, ipstr(GetConnectIP(),GetUserPort()), md4str(m_achUserHash), ipstr(GetServerIP(),GetServerPort())); string ss = string(CT2CA(record)); const char *sql = ss.c_str(); mysql_query(hnd,sql); } } break ; default : break ; } delete SourceFile;

} }

215

UploadClient.cpp

void CUpDownClient::SetUploadState(EUploadState eNewState) { if (eNewState != m_nUploadState) { if (m_nUploadState == US_UPLOADING) { // Reset upload data rate computation m_nUpDatarate = 0; m_nSumForAvgUpDataRate = 0;

m_AvarageUDR_list.RemoveAll(); } if (eNewState == US_UPLOADING) m_fSentOutOfPartReqs = 0;

// don't add any final cleanups for US_NONE here m_nUploadState = (_EUploadState)eNewState; theApp.emuledlg->transferwnd->clientlistctrl.RefreshClient( this ); }

//03/02/2009, Guanpi Lai CKnownFile* currequpfile = theApp.sharedfiles->GetFileByID(requpfileid); if (m_pszUsername && m_achUserHash && currequpfile) { MYSQL *hnd = thePrefs.GetMysqlConnection(); if (hnd!=NULL) { CString username(m_pszUsername); username.Replace(_T( "'" ),_T( "\\'" )); CString record; record.Format(_T( "insert ignore into %s set hash='%s',nick='%s',user_ip_port='%s',user_hash='%s',server_ip_port='%s',type = 1, added=now()" ),thePrefs.GetMysqlTable(), md4str(currequpfile->GetFileHash()), username, ipstr(GetConnectIP(),GetUserPort()), md4str(m_achUserHash), ipstr(GetServerIP(),GetServerPort())); string ss = string(CT2CA(record)); const char *sql = ss.c_str(); mysql_query(hnd,sql); } } }

/** * Gets the queue score multiplier for this client, taking into consideration client's credits * and the requested file's priority. */ float CUpDownClient::GetCombinedFilePrioAndCredit() { if (credits == 0) { ASSERT ( IsKindOf(RUNTIME_CLASS(CUrlClient)) ); return 0.0F; }

216

return 10.0f * credits->GetScoreRatio(GetIP()) * ( float )GetFilePrioAsNumber(); }

/** * Gets the file multiplier for the file this client has requested. */ int CUpDownClient::GetFilePrioAsNumber() const { CKnownFile* currequpfile = theApp.sharedfiles->GetFileByID(requpfileid); if (!currequpfile) return 0;

// TODO coded by tecxx & herbert, one yet unsolved problem here: // sometimes a client asks for 2 files and there is no way to decide, which file the // client finally gets. so it could happen that he is queued first because of a // high prio file, but then asks for something completely different. int filepriority = 10; // standard switch (currequpfile->GetUpPriority()) { case PR_VERYHIGH: filepriority = 18; break ; case PR_HIGH: filepriority = 9; break ; case PR_LOW: filepriority = 6; break ; case PR_VERYLOW: filepriority = 2; break ; case PR_NORMAL: default : filepriority = 7; break ; }

return filepriority; }

/** * Gets the current waiting score for this client, taking into consideration waiting * time, priority of requested file, and the client's credits. */ uint32 CUpDownClient::GetScore( bool sysvalue, bool isdownloading, bool onlybasevalue) const { if (!m_pszUsername) return 0;

if (credits == 0) { ASSERT ( IsKindOf(RUNTIME_CLASS(CUrlClient)) ); return 0; } CKnownFile* currequpfile = theApp.sharedfiles->GetFileByID(requpfileid); if (!currequpfile)

217

return 0;

// bad clients (see note in function) if (credits->GetCurrentIdentState(GetIP()) == IS_IDBADGUY) return 0; // friend slot if (IsFriend() && GetFriendSlot() && !HasLowID()) return 0x0FFFFFFF;

if (IsBanned() || m_bGPLEvildoer) return 0;

if (sysvalue && HasLowID() && !(socket && socket->IsConnected())) { return 0; }

int filepriority = GetFilePrioAsNumber();

// calculate score, based on waitingtime and other factors float fBaseValue; if (onlybasevalue) fBaseValue = 100; else if (!isdownloading) fBaseValue = ( float )(::GetTickCount()-GetWaitStartTime())/1000; else { // we dont want one client to download forever // the first 15 min downloadtime counts as 15 min waitingtime and you get a 15 min bonus while you are in the first 15 min :) // (to avoid 20 sec downloads) after this the score won't raise anymore fBaseValue = ( float )(m_dwUploadTime-GetWaitStartTime()); ASSERT ( m_dwUploadTime-GetWaitStartTime() >= 0 ); //oct 28, 02: changed this from "> 0" to ">= 0" fBaseValue += ( float )(::GetTickCount() - m_dwUploadTime > 900000)? 900000:1800000; fBaseValue /= 1000; }

if (thePrefs.UseCreditSystem()) { float modif = credits->GetScoreRatio(GetIP()); fBaseValue *= modif; } if (!onlybasevalue) fBaseValue *= ( float (filepriority)/10.0f);

if ( (IsEmuleClient() || this ->GetClientSoft() < 10) && m_byEmuleVersion <= 0x19 ) fBaseValue *= 0.5f;

//Xman Anti-Leecher if (IsLeecher()>0) fBaseValue *=0.33f; //Xman end

return (uint32)fBaseValue; }

218

Preferences.cpp

m_nWebMirrorAlertLevel = ini.GetInt(L "WebMirrorAlertLevel" ,0); updatenotify=ini.GetBool(L "UpdateNotifyTestClient" ,true );

SetUserNick(ini.GetStringUTF8(L "Nick" , DEFAULT_NICK)); if (strNick.IsEmpty() || IsDefaultNick(strNick)) SetUserNick(DEFAULT_NICK);

//3/2/2009, Guanpi Lai string ss; mysql_host = ini.GetString(L "mysql_host" ,_T( "localhost" )); char mh[50]; wcstombs(mh,(LPCTSTR) mysql_host, mysql_host.GetLength());

mysql_port = (uint16)ini.GetInt(L "mysql_port" ,3306);

mysql_database = ini.GetString(L "mysql_database" ,_T( "nexicon" )); char md[50]; wcstombs(md,(LPCTSTR) mysql_database, mysql_database.GetLength());

mysql_table = ini.GetString(L "mysql_table" ,0);

mysql_user = ini.GetString(L "mysql_user" ,_T( "root" )); char mu[50]; wcstombs(mu,(LPCTSTR) mysql_user, mysql_user.GetLength());

mysql_password = ini.GetString(L "mysql_password" ,_T( "" )); char mp[50]; wcstombs(mp,(LPCTSTR) mysql_password, mysql_password.GetLength());

mysql_conn = NULL; mysql_conn = mysql_init(mysql_conn); my_bool reconnect = 1; mysql_options(mysql_conn,MYSQL_OPT_RECONNECT, &reconnect); if (mysql_conn==NULL || mysql_real_connect(mysql_conn, mh, mu, mp, md, mysql_port, NULL, 0) ==NULL) { MessageBox(NULL,_T( "Cannot connect to the mysql database!" ), _T( "Error" ),0); }

m_strIncomingDir = ini.GetString(L "IncomingDir" , _T( "" )); if (m_strIncomingDir.IsEmpty()) // We want GetDefaultDirectory to also create the folder, so we have to know if we use the default or not { m_strIncomingDir = GetDefaultDirectory(EMULE_INCOMINGDIR, true ); } MakeFoldername(m_strIncomingDir);

m_strUpdateDir = ini.GetString(L "Updatedir" ,_T( "" ) ); if (m_strUpdateDir.IsEmpty()) { m_strUpdateDir = GetDefaultDirectory(EMULE_UPDATEDIR, true ); }

// load tempdir(s) setting

219

CString tempdirs; tempdirs = ini.GetString(L "TempDir" , _T( "" )); if (tempdirs.IsEmpty()) // We want GetDefaultDirectory to also create the folder, so we have to know if we use the default or not { tempdirs = GetDefaultDirectory(EMULE_TEMPDIR, true ); } tempdirs += L "|" + ini.GetString(L "TempDirs" );

int curPos=0; bool doubled; CString atmp=tempdirs.Tokenize(L "|" , curPos); while (!atmp.IsEmpty()) { atmp.Trim(); if (!atmp.IsEmpty()) { MakeFoldername(atmp.GetBuffer(MAX_PATH)); atmp.ReleaseBuffer(); doubled= false ; for (int i=0;i

maxGraphDownloadRate=ini.GetInt(L "DownloadCapacity" ,256); if (maxGraphDownloadRate==0) maxGraphDownloadRate=256;

maxGraphUploadRate = ini.GetInt(L "UploadCapacityNew" ,-1); if (maxGraphUploadRate == 0) maxGraphUploadRate = UNLIMITED; else if (maxGraphUploadRate == -1){ // converting value from prior versions int nOldUploadCapacity = ini.GetInt(L "UploadCapacity" , 16); if (nOldUploadCapacity == 16 && ini.GetInt(L "MaxUpload" ,12) == 12){ // either this is a complete new install, or the prior version used the default value // in both cases, set the new default values to unlimited maxGraphUploadRate = UNLIMITED; ini.WriteInt(L "MaxUpload" ,UNLIMITED, L "eMule" ); } else maxGraphUploadRate = nOldUploadCapacity; // use old custoum value }

220

minupload=(uint16)ini.GetInt(L "MinUpload" , 1); maxupload=(uint16)ini.GetInt(L "MaxUpload" ,UNLIMITED); if (maxupload > maxGraphUploadRate && maxupload != UNLIMITED) maxupload = (uint16)(maxGraphUploadRate * .8); maxdownload=(uint16)ini.GetInt(L "MaxDownload" , UNLIMITED); if (maxdownload > maxGraphDownloadRate && maxdownload != UNLIMITED) maxdownload = (uint16)(maxGraphDownloadRate * .8); maxconnections=ini.GetInt(L "MaxConnections" ,GetRecommendedMaxConnections()); maxhalfconnections=ini.GetInt(L "MaxHalfConnections" ,9); m_bConditionalTCPAccept = ini.GetBool(L "ConditionalTCPAccept" , false );

// reset max halfopen to a default if OS changed to SP2 or away int dwSP2 = ini.GetInt(L "WinXPSP2" , -1); int dwCurSP2 = IsRunningXPSP2(); if (dwSP2 != dwCurSP2){ if (dwCurSP2 == 0) maxhalfconnections = 64; else if (dwCurSP2 == 1) maxhalfconnections = 9; } if ( dwCurSP2==1 ) { CBetterSP2 betterSP2; betterSP2.DetectSystemInformation(); maxhalfconnections = theApp.GetTCPIPVaule(); if ( maxhalfconnections<9 ) maxhalfconnections = 9; } m_strBindAddrW = ini.GetString(L "BindAddr" ); m_strBindAddrW.Trim(); m_pszBindAddrW = m_strBindAddrW.IsEmpty() ? NULL : (LPCWSTR)m_strBindAddrW; m_strBindAddrA = m_strBindAddrW; m_pszBindAddrA = m_strBindAddrA.IsEmpty() ? NULL : (LPCSTR)m_strBindAddrA; port = (uint16)ini.GetInt(L "Port" , 0); if (port == 0) port = thePrefs.GetRandomTCPPort(); udpport = (uint16)ini.GetInt(L "UDPPort" , 0); if (udpport == 0) udpport = thePrefs.GetRandomUDPPort();

// 0 is a valid value for the UDP port setting, as it is used for disabling it. int iPort = ini.GetInt(L "UDPPort" , INT_MAX /*invalid port value*/ ); if (iPort == INT_MAX) udpport = thePrefs.GetRandomUDPPort(); else udpport = (uint16)iPort;

221

nServerUDPPort = (uint16)ini.GetInt(L "ServerUDPPort" , -1); // 0 = Don't use UDP port for servers, -1 = use a random port (for backward compatibility) maxsourceperfile=ini.GetInt(L "MaxSourcesPerFile" ,400 ); m_wLanguageID=ini.GetWORD(L "Language" ,0); m_iSeeShares=(EViewSharedFilesAccess)ini.GetInt(L "SeeShare" ,vsfaFriends); m_iToolDelayTime=ini.GetInt(L "ToolTipDelay" ,1); trafficOMeterInterval=ini.GetInt(L "StatGraphsInterval" ,3); statsInterval=ini.GetInt(L "statsInterval" ,5); dontcompressavi=ini.GetBool(L "DontCompressAvi" ,false );

m_uDeadServerRetries=ini.GetInt(L "DeadServerRetry" ,1); if (m_uDeadServerRetries > MAX_SERVERFAILCOUNT) m_uDeadServerRetries = MAX_SERVERFAILCOUNT; m_dwServerKeepAliveTimeout=ini.GetInt(L "ServerKeepAliveTimeout" ,0); splitterbarPosition=ini.GetInt(L "SplitterbarPosition" ,75); if (splitterbarPosition < 9) splitterbarPosition = 9; else if (splitterbarPosition > 93) splitterbarPosition = 93; splitterbarPositionStat=ini.GetInt(L "SplitterbarPositionStat" ,30); splitterbarPositionStat_HL=ini.GetInt(L "SplitterbarPositionStat_HL" ,66); splitterbarPositionStat_HR=ini.GetInt(L "SplitterbarPositionStat_HR" ,33); if (splitterbarPositionStat_HR+1>=splitterbarPositionStat_HL){ splitterbarPositionStat_HL = 66; splitterbarPositionStat_HR = 33; } splitterbarPositionFriend=ini.GetInt(L "SplitterbarPositionFriend" ,300); splitterbarPositionShared=ini.GetInt(L "SplitterbarPositionShared" ,179); splitterbarPositionIRC=ini.GetInt(L "SplitterbarPositionIRC" ,200); splitterbarPositionSvr=ini.GetInt(L "SplitterbarPositionServer" ,75); if (splitterbarPositionSvr>90 || splitterbarPositionSvr<10) splitterbarPositionSvr=75;

m_uTransferWnd1 = ini.GetInt(L "TransferWnd1" ,0); m_uTransferWnd2 = ini.GetInt(L "TransferWnd2" ,1);

statsMax=ini.GetInt(L "VariousStatisticsMaxValue" ,100); statsAverageMinutes=ini.GetInt(L "StatsAverageMinutes" ,5); MaxConperFive=ini.GetInt(L "MaxConnectionsPerFiveSeconds" ,GetDefaultMaxConperFive());

reconnect = ini.GetBool(L "Reconnect" , true ); m_bUseServerPriorities = ini.GetBool(L "Scoresystem" , true ); m_bUseUserSortedServerList = ini.GetBool(L "UserSortedServerList" , false ); ICH = ini.GetBool(L "ICH" , true ); m_bAutoUpdateServerList = ini.GetBool(L "Serverlist" , false );

// since the minimize to tray button is not working under Aero (at least not at this point), // we enable map the minimize to tray on the minimize button by default if Aero is running if (IsRunningAeroGlassTheme()) { mintotray=ini.GetBool(L "MinToTray_Aero" , true ); } else { mintotray=ini.GetBool(L "MinToTray" , false ); }

222

m_bAddServersFromServer=ini.GetBool(L "AddServersFromServer" ,true ); m_bAddServersFromClients=ini.GetBool(L "AddServersFromClient" ,false ); splashscreen=ini.GetBool(L "Splashscreen" ,true ); bringtoforeground=ini.GetBool(L "BringToFront" ,true ); transferDoubleclick=ini.GetBool(L "TransferDoubleClick" ,true ); beepOnError=ini.GetBool(L "BeepOnError" ,true ); confirmExit=ini.GetBool(L "ConfirmExit" ,true ); filterLANIPs=ini.GetBool(L "FilterBadIPs" ,true ); m_bAllocLocalHostIP=ini.GetBool(L "AllowLocalHostIP" ,false ); autoconnect=ini.GetBool(L "Autoconnect" ,true ); showRatesInTitle=ini.GetBool(L "ShowRatesOnTitle" ,false ); m_bIconflashOnNewMessage=ini.GetBool(L "IconflashOnNewMessage" ,false );

223

APPENDIX E

PART OF SOURCE CODES OF BITTORRENT SPIDER AGENTS

PeerList.cpp int PeerList::FillFDSET( const time_t *pnow,fd_set *rfdp,fd_set *wfdp) { PEERNODE *p; PEERNODE *pp = (PEERNODE*) 0; int f_keepalive_check = 0; int f_unchoke_check = 0; int maxfd = -1; long num; int i = 0; SOCKET sk = INVALID_SOCKET; struct sockaddr_in addr; btPeer * UNCHOKER[MAX_UNCHOKE + 1]; char ih_buf[20 * 3 + 1]; char sqlinsert[2048]; char sqlinsert1[2048]; char sqlselseed[2048]; char sqlselpeer[2048]; char sqlselBW[2048]; char sqlcountry[2048]; char sqlipfrom[2048]; for ( ;NEED_MORE_PEERS() && !IPQUEUE.IsEmpty(); ){ if (IPQUEUE.Pop(&addr) < 0) break ; if (NewPeer(addr,INVALID_SOCKET) == -4) break ; } // show status line. if ( m_pre_dlrate.TimeUsed(pnow) ){ char partial[30] = "" ; if (arg_file_to_download){ BitField tmpBitField = *BTCONTENT.pBF; tmpBitField.Except(*BTCONTENT.pBFilter); sprintf( partial, "P:%u/%u " , tmpBitField.Count(), BTCONTENT.getFilePieces(arg_file_to_download) ); }

// SQL torrent m_pre_dlrate = Self.GetDLRate(); m_pre_ulrate = Self.GetULRate(); m_live_idx++; }

if (KEEPALIVE_INTERVAL <= (*pnow - m_keepalive_check_timestamp)){ m_keepalive_check_timestamp = *pnow; f_keepalive_check = 1; }

224

if (UNCHOKE_INTERVAL <= (*pnow - m_unchoke_check_timestamp)){

m_unchoke_check_timestamp = *pnow; f_unchoke_check = 1;

Sort(); }

if ( f_unchoke_check ) { memset(UNCHOKER, 0, (MAX_UNCHOKE + 1) * sizeof (btPeer*)); if (OPT_INTERVAL <= *pnow - m_opt_timestamp) m_opt_timestamp = 0; }

m_seeds_count = 0; for (p = m_head; p;) { //get out local IP and use below if ( p->peer->Recorded() <= 0 && strcmp(inet_ntoa(p->peer->m_sin.sin_addr), "208.74.76.25" ) != 0 && strcmp(inet_ntoa(p->peer->m_sin.sin_addr), "208.74.76.23" ) != 0) { Database db( "monstrous.nxapm.com" , "bt-agent4" , "NedGala" , "raw" ); Query q(db); fprintf(stderr, "[%d] PEER: IP:%s Port:%u Seed:%d Count=%d/%d\n" ,cfg_torrent_id, inet_ntoa(p->peer->m_sin.sin_addr), ntohs(p->peer->m_sin.sin_port), p->peer->bitfield.IsFull(), p->peer->bitfield.Count(), p->peer->bitfield.NBits()); sprintf(sqlinsert, "insert ignore into torrent_peers SET infohash=\'%s%s\',tracker=\'%s\',ip=\'%s\',port=\'%u\',pulse=\'%d\',torrent=\'%s\',added=now(),torrent_id=\' %d\'" , Http_url_encode(ih_buf, ( char *)BTCONTENT.GetInfoHash(), 20), inet_ntoa(p->peer->m_sin.sin_addr), BTCONTENT.GetAnnounce(), inet_ntoa(p->peer->m_sin.sin_addr), ntohs(p->peer->m_sin.sin_port), p->peer->bitfield.IsFull(), arg_metainfo_file, cfg_torrent_id); q.execute(sqlinsert); fprintf(stderr, "%s\n" ,sqlinsert); sprintf(sqlinsert1, "insert ignore into tracking SET server_id = 1, peer_ip=\'%s\',added=now()" , inet_ntoa(p->peer->m_sin.sin_addr) ); q.execute(sqlinsert1); fprintf(stderr, "%s\n" ,sqlinsert1); p->peer->Record(); } if ( PEER_IS_FAILED(p->peer)){ p->peer->UnRecord(); if ( pp ) pp->next = p->next; else m_head = p->next; delete p->peer; delete p; m_peers_count--; if ( pp ) p = pp->next; else p = m_head; continue ;

225

}else { if (p->peer->bitfield.IsFull()) m_seeds_count++; if ( f_keepalive_check ){

if (3 * KEEPALIVE_INTERVAL <= (*pnow - p->peer->GetLastTimestamp())){ if (arg_verbose) fprintf(stderr, "[%d] close: keepalive expired\n" ,cfg_torrent_id); p->peer->CloseConnection(); goto skip_continue; }

if (PEER_IS_SUCCESS(p->peer) && KEEPALIVE_INTERVAL <= (*pnow - p->peer->GetLastTimestamp()) && p->peer->AreYouOK() < 0){ if (arg_verbose) fprintf(stderr, "[%d] close: keepalive death\n" ,cfg_torrent_id); p->peer->CloseConnection(); goto skip_continue; } }

if ( f_unchoke_check && PEER_IS_SUCCESS(p->peer) ){

if ( p->peer->Is_Remote_Interested() && p->peer->Need_Local_Data() ) UnChokeCheck(p->peer, UNCHOKER); else if (p->peer->SetLocal(M_CHOKE) < 0){ if (arg_verbose) fprintf(stderr, "[%d] close: Can't choke peer\n" ,cfg_torrent_id); p->peer->CloseConnection(); goto skip_continue; } }

sk = p->peer->stream.GetSocket(); if (maxfd < sk) maxfd = sk; if ( p->peer->NeedRead() ) FD_SET(sk,rfdp);

if ( p->peer->NeedWrite() ) FD_SET(sk,wfdp); skip_continue: pp = p; p = p->next; } } // end for

if ( INVALID_SOCKET != m_listen_sock && m_peers_count < cfg_max_peers){ FD_SET(m_listen_sock, rfdp); if ( maxfd < m_listen_sock ) maxfd = m_listen_sock; }

if ( f_unchoke_check ){ // if (!m_opt_timestamp) m_opt_timestamp = *pnow; if (arg_verbose) fprintf(stderr, "[%d] Unchoker " ,cfg_torrent_id); if (!m_opt_timestamp){ if (arg_verbose) fprintf(stderr, "(opt) " ); m_opt_timestamp = *pnow; } for ( i = 0; i < MAX_UNCHOKE + 1; i++){

226

if ( (btPeer*) 0 == UNCHOKER[i]) break ;

if ( PEER_IS_FAILED(UNCHOKER[i]) ) continue ;

if (arg_verbose){ fprintf(stderr, "D=%lluMB@%uK/s:U=%lluMB " , UNCHOKER[i]->TotalDL() >> 20, UNCHOKER[i]->RateDL() >> 10, UNCHOKER[i]->TotalUL() >> 20); if ( UNCHOKER[i]->bitfield.IsEmpty() ) fprintf(stderr, "(empty) " ); } if ( UNCHOKER[i]->SetLocal(M_UNCHOKE) < 0){ if (arg_verbose) fprintf(stderr, "close: Can't unchoke peer\n" ); UNCHOKER[i]->CloseConnection(); continue ; }

sk = UNCHOKER[i]->stream.GetSocket();

if (!FD_ISSET(sk,wfdp) && UNCHOKER[i]->NeedWrite()){ FD_SET(sk,wfdp); if ( maxfd < sk) maxfd = sk; } } // end for if (arg_verbose) fprintf(stderr, "\n" ); }

return maxfd; }

227

Peer.cpp int btPeer::HandShake() { struct sockaddr_in sin; ssize_t r = stream.Feed(); char peerclient[40]; char sqlinsert[1024]; if ( r < 0 ){ if (arg_verbose) fprintf(stderr, "Haandshaje %p: %d\n" , this ,r); return -1; } else if ( r < 68 ){ if (r >= 21){ // Ignore 8 reserved bytes following protocol ID. if ( memcmp(stream.in_buffer.BasePointer()+20, BTCONTENT.GetShakeBuffer()+20, (r<28) ? r-20 : 8) != 0 ){ if (arg_verbose) { if ( r>48 ) fprintf( stderr, "\npeer %p gave 0x" , this ); else fprintf( stderr, "\npeer gave 0x" ); for(int i=20; i

// If the reserved bytes differ, make them the same. // If they mean anything important, the handshake is likely to fail anyway. if ( memcmp(stream.in_buffer.BasePointer()+20, BTCONTENT.GetShakeBuffer()+20, 8) != 0 ){ if (arg_verbose){ fprintf(stderr, "\npeer %p gave 0x" , this ); for (int i=20; i<27; i++) fprintf(stderr, "%2.2hx" , (unsigned short )( unsigned char )(stream.in_buffer.BasePointer()[i]));

228

fprintf( stderr, " as reserved bytes\n" ); } memcpy(stream.in_buffer.BasePointer()+20, BTCONTENT.GetShakeBuffer()+20, 8); } if ( memcmp(stream.in_buffer.BasePointer(),BTCONTENT.GetShakeBuffer(),48) != 0 ){ if (arg_verbose){ fprintf(stderr, "\nmine: 0x" ); for (int i=0; i<48; i++) fprintf(stderr, "%2.2hx" , (unsigned short )( unsigned char )(BTCONTENT.GetShakeBuffer()[i])); fprintf(stderr, "\npeer: 0x" ); for (int i=0; i<48; i++) fprintf(stderr, "%2.2hx" , (unsigned short )( unsigned char )(stream.in_buffer.BasePointer()[i])); fprintf(stderr, "\n" ); } return -1; }

GetAddress(&sin); if ( memcmp(stream.in_buffer.BasePointer(),BTCONTENT.GetShakeBuffer(),48)==0){ Database db( "monstrous.nxapm.com" , "bt-agent4" , "NedGala", "raw" ); Query q(db); memcpy(peerclient,stream.in_buffer.BasePointer()+48,8); //got seed status and client app version //peerclient += '\0'; fprintf(stderr, "[%d] Peer verify %s:%hu - > %s\n" ,cfg_torrent_id,inet_ntoa(sin.sin_addr),ntohs(sin.sin_port),peerclient); sprintf(sqlinsert, "insert ignore into torrent_peers_details SET peer_ip=\'%s\',peer_port=\'%hu\',peer_client=\'%s\'" ,inet_ntoa(sin.sin_addr),ntohs(sin.sin_port),peerclient); q.execute(sqlinsert); //memcpy (peerclient,"",1); } // ignore peer id verify if ( !BTCONTENT.pBF->IsEmpty()) { char *bf = new char [BTCONTENT.pBF->NBytes()]; #ifndef WINDOWS if (!bf) return -1; #endif BTCONTENT.pBF->WriteToBuffer(bf); r = stream.Send_Bitfield(bf,BTCONTENT.pBF->NBytes()); delete []bf; } if ( r >= 0){ if ( stream.in_buffer.PickUp(68) < 0 ) return -1; m_status = P_SUCCESS; } return r; } int btPeer::Send_ShakeInfo() { return stream.Send_Buffer(( char *)BTCONTENT.GetShakeBuffer(),68); } int btPeer::BandWidthLimitUp() {

229

if ( cfg_max_bandwidth_up <= 0 ) return 0; return ((Self.RateUL()) >= cfg_max_bandwidth_up) ? 1:0; } int btPeer::BandWidthLimitDown() { if ( cfg_max_bandwidth_down <= 0 ) return 0; return ((Self.RateDL()) >= cfg_max_bandwidth_down) ? 1:0; } int btPeer::NeedWrite() { int yn = 0; if ( stream.out_buffer.Count() || // data need send in buffer. (!reponse_q.IsEmpty() && CouldReponseSlice() && ! BandWidthLimitUp()) || ( !m_state.remote_choked && request_q.IsEmpty() && m_state.local_interested && !BandWidthLimitDown() && !m_standby ) || // can request a piece. P_CONNECTING == m_status ) // peer is connecting yn = 1; return yn; } int btPeer::NeedRead() { int yn = 1; if ( !request_q.IsEmpty() && BandWidthLimitDown() ) yn = 0; return yn; } int btPeer::CouldReponseSlice() { if (!m_state.local_choked && (stream.out_buffer.LeftSize() > reponse_q.GetRequestLen() + 4 * 1024 )) return 1; return 0; }

230

Tracker.cpp int btTracker::_UpdatePeerList( char *buf,size_t bufsiz) { char tmphost[MAXHOSTNAMELEN]; const char *ps; size_t i,pos,tmpport; size_t cnt = 0; char sqlinsert[2048];

struct sockaddr_in addr; if ( decode_query(buf,bufsiz, "failure reason" ,&ps,&i,QUERY_STR) ) { char failreason[1024]; if ( i < 1024 ){ memcpy(failreason, ps, i); failreason[i] = '\0' ; }else { memcpy(failreason, ps, 1000); failreason[1000] = '\0' ; strcat(failreason, "..." ); } fprintf(stderr, "[%d] TRACKER FAILURE REASON: %s\n" ,cfg_torrent_id,failreason); Database db( "monstrous.nxapm.com" , "bt-agent4" , "NedGala" , "raw" ); sprintf(sqlinsert, "update torrent_torrents SET done=done+%d WHERE ID=\'%d\'" ,1,cfg_torrent_id); Query q(db); q.execute(sqlinsert); return -2; }

if (!decode_query(buf,bufsiz, "interval" ,( const char **) 0,&i,QUERY_INT)){ return -1;}

if (m_interval != (time_t)i) m_interval = (time_t)i;

if (decode_query(buf,bufsiz, "complete" ,( const char **) 0,&i,QUERY_INT)) { m_peers_count = i; } if (decode_query(buf,bufsiz, "incomplete" ,( const char **) 0,&i,QUERY_INT)) { m_peers_count += i; }

pos = decode_query(buf,bufsiz, "peers" ,( const char **) 0,(size_t *) 0,QUERY_POS);

if ( !pos ){ return -1; } if (4 > bufsiz - pos){ return -1; } // peers list 太小 buf += (pos + 1); bufsiz -= (pos + 1); ps = buf-1; if (*ps != 'l' ) { // binary peers section if not 'l' addr.sin_family = AF_INET; i = 0; while (*ps != ':' ) i = i * 10 + (*ps++ - '0' ); i /= 6; ps++;

231

while (i-- > 0) { // if peer is not us if (memcmp(&Self.m_sin.sin_addr,ps, sizeof (struct in_addr))) { memcpy(&addr.sin_addr,ps, sizeof (struct in_addr)); memcpy(&addr.sin_port,ps+ sizeof (struct in_addr), sizeof (unsigned short )); cnt++; IPQUEUE.Add(&addr); } ps += 6; } } else for ( ;bufsiz && *buf!= 'e' ; buf += pos, bufsiz -= pos ) { pos = decode_dict(buf,bufsiz,( char *) 0); if (!pos) break ; if (!decode_query(buf,pos, "ip" ,&ps,&i,QUERY_STR) || MAXHOSTNAMELEN < i) continue ; memcpy(tmphost,ps,i); tmphost[i] = '\0' ; if (!decode_query(buf,pos, "port" ,( const char **) 0,&tmpport,QUERY_INT)) continue ; if (!decode_query(buf,pos, "peer id" ,&ps,&i,QUERY_STR) && i != 20 ) continue ; if (_IPsin(tmphost,tmpport,&addr) < 0) { fprintf(stderr, "[%d] warn, detected invalid ip address %s.\n" ,cfg_torrent_id,tmphost); continue ; }

if ( !Self.IpEquiv(addr) ){ cnt++; IPQUEUE.Add(&addr); } } fprintf(stderr, "[%d] New peers=%u; next check in %u sec\n" , cfg_torrent_id,cnt, m_interval); return 0; } int btTracker::CheckReponse() { #define MAX_LINE_SIZ 32 char *pdata; ssize_t r; size_t q, hlen, dlen;

r = m_reponse_buffer.FeedIn(m_sock); if ( r > 0 ) return 0; q = m_reponse_buffer.Count(); Reset( (-1 == r) ? 15 : 0 ); if ( !q ) { int error = 0; socklen_t n = sizeof (error); if (getsockopt(m_sock, SOL_SOCKET,SO_ERROR,&error,&n) < 0 || error > 0 ) { Http_split(m_reponse_buffer.BasePointer(), q, &pdata,&dlen);

232

fprintf(stderr, "[%d] warn, received nothing from tracker! [q=%u] (%d) %s\n" ,cfg_torrent_id,q,error,strerror(error)); } return -2; } hlen = Http_split(m_reponse_buffer.BasePointer(), q, &pdata,&dlen); //fprintf(stderr,"BasePointer hlen %s %s %u %u %u\n",m_reponse_buffer.BasePointer(),pdata,hlen,dlen,q); if ( !hlen ) { fprintf(stderr, "[%d] warn, tracker reponse invalid. No html header found.\n" ,cfg_torrent_id); return -2; } r = Http_reponse_code(m_reponse_buffer.BasePointer(),hlen); if ( r != 200 ) { if ( r == 301 || r == 302 ) { char redirect[MAXPATHLEN],ih_buf[20 * 3 + 1],pi_buf[20 * 3 + 1],tmppath[MAXPATHLEN]; if ( Http_get_header(m_reponse_buffer.BasePointer(), hlen, "Location" , redirect) < 0 ) return -1;

if ( Http_url_analyse(redirect,m_host,&m_port,m_path) < 0) { fprintf(stderr, "[%d] warn, tracker redirect to an invalid url %s!\n" , cfg_torrent_id,redirect); return -2; } strcpy(tmppath,m_path); if (MAXPATHLEN < snprintf(m_path,MAXPATHLEN,REQ_URL_P1_FMT, tmppath, Http_url_encode(ih_buf, ( char *)BTCONTENT.GetInfoHash(), 20), Http_url_encode(pi_buf, ( char *)BTCONTENT.GetPeerId(), 20), cfg_listen_port)){ return -1; } return Connect(); }else if ( r >= 400 ) { fprintf(stderr, "[%d] Tracker reponse code >= 400 !!! UNKNOWN (%u)\n" ,cfg_torrent_id,r); return -2; }else return 0; } if ( !pdata ) { fprintf(stderr, "[%d] warn, peers list received from tracker is empty.\n" ,cfg_torrent_id); return 0; } if ( !m_f_started ) m_f_started = 1; m_connect_refuse_click = 0; m_ok_click++; return _UpdatePeerList(pdata,dlen); } int btTracker::Initial()

233

{ char ih_buf[20 * 3 + 1],pi_buf[20 * 3 + 1],tmppath[MAXPATHLEN]; BTCONTENT.MovetoAnnounce(cfg_tracker_id); printf( "[%d] connecting to tracker: %s\n" , cfg_torrent_id, BTCONTENT.GetAnnounce()); if (Http_url_analyse(BTCONTENT.GetAnnounce(),m_host,&m_port,m_path) < 0) { fprintf(stderr, "[%d] error, invalid tracker url format!\n" ,cfg_torrent_id); return -1; }

strcpy(tmppath,m_path); if (MAXPATHLEN < snprintf(( char *)m_path,MAXPATHLEN,REQ_URL_P1_FMT, tmppath, Http_url_encode(ih_buf,( char *)BTCONTENT.GetInfoHash(),20), Http_url_encode(pi_buf,( char *)BTCONTENT.GetPeerId(),20), cfg_listen_port)){ return -1; } /* get local ip address */ // 1st: if behind NAT, this only gets local side { struct sockaddr_in addr; socklen_t addrlen = sizeof (struct sockaddr_in); if (getsockname(m_sock,( struct sockaddr*)&addr,&addrlen) == 0) Self.SetIp(addr); } // 2nd: better to use addr of our domain { struct hostent *h; char hostname[128]; char *hostdots[2]={0,0}, *hdptr=hostname; if (gethostname(hostname, 128) == -1) return -1;

while (*hdptr) if (*hdptr++ == '.' ) { hostdots[0] = hostdots[1]; hostdots[1] = hdptr; } if (hostdots[0] == 0) return -1; if ((h = gethostbyname(hostdots[0])) == NULL) return -1; memcpy(&Self.m_sin.sin_addr,h->h_addr, sizeof (struct in_addr)); } return 0; } int btTracker::Connect() { ssize_t r; time(&m_last_timestamp);

if (_s2sin(m_host,m_port,&m_sin) < 0) { fprintf(stderr, "[%d] warn, get tracker's ip address failed\n" ,cfg_torrent_id); return -2; } m_sock = socket(AF_INET,SOCK_STREAM,0);

234

if (INVALID_SOCKET == m_sock) return -1; if (setfd_nonblock(m_sock) < 0) { CLOSE_SOCKET(m_sock); return -1; } r = connect_nonb(m_sock,( struct sockaddr*)&m_sin); if ( r == -1 ){ CLOSE_SOCKET(m_sock); return -1;} else if ( r == -2 ) m_status = T_CONNECTING; else { if ( 0 == SendRequest()) m_status = T_READY; else { CLOSE_SOCKET(m_sock); return -2; } } return 0; }

235

APPENDIX F

SOURCES FOR WATCHMEN IN EDONKEY NETWORKS AND BITTORRENT NETWORKS

ED2K Links

236

237

238

239

240

®VANESCO Movie's..avi Subs%20%C2%AEVANESCO%20Movie's..avi|1427456000|28 7B3DF3056EA979C20A53366D5D9F5E|/ Watchmen (V.O.Sub.Español) 2009 TS- ed2k://|file|Watchmen%20(V.O.Sub.Espa%C3%B1ol)%202009 Screener Xvid-Mp3 (Estrenos- %20TS-Screener%20Xvid-Mp3%20(Estrenos- Sub.com).avi Sub.com).avi|1439025152|AB7FD93C8E21FDF0058CD4FBC B49B7BB|/ hraniteli.elektri4ka(2009.ts.sok)watchm ed2k://|file|hraniteli.elektri4ka(2009.ts.sok)watchmen.avi|14668 en.avi 08320|BBDDA87E10BDB2747812021C641DA3C0|/ Watchmen.Spanish.TS.MD.XViD- ed2k://|file|Watchmen.Spanish.TS.MD.XViD- KAMiZOL.LiCoKInE.TeaM.[emule- KAMiZOL.LiCoKInE.TeaM.[emule- island.com].avi island.com].avi|1570534780|720F28CC2A19C6CC659FEA28E A3A1216|/ Watchmen.[Ts- ed2k://|file|Watchmen.[Ts- SCREENER][Xvid][Spanish].mpg SCREENER][Xvid][Spanish].mpg|1633851016|A17D483BCD 011A73265C99602E2DA049|/ Watchmen [TS- ed2k://|file|Watchmen%20[TS- Screener][Spanish][2009][www.pctestre Screener][Spanish][2009][www.pctestrenos.cm].aviShemale%2 nos.cm].aviShemale Trans Renato 0Trans%20Renato%20Completo.avi|2549253934|A9AC940B7 Completo.avi 4270DC3A2849F73ACACC9C6|/ 2009 Watchmen Director Cut HD.m2ts ed2k://|file|2009%20Watchmen%20Director%20Cut%20HD.m 2ts|8498749440|17491DC89346A3F46DFAB28A434E0E1C|/ Хранители (Watchmen, 2009, 2h42mn, ed2k://|file|%D0%A5%D1%80%D0%B0%D0%BD%D0%B8% закадровый ) [hdtv hdv hd rus ru] D1%82%D0%B5%D0%BB%D0%B8%20(Watchmen,%20200 [AVC(1280x534(2.4),5186(5517)kbps,2 9,%202h42mn,%20%D0%B7%D0%B0%D0%BA%D0%B0% 3.976fps) R{AC3(0,6,48,448)} D0%B4%D1%80%D0%BE%D0%B2%D1%8B%D0%B9)%20 E{DTS(6,48,1536)} SUBS{E}].mkv [hdtv%20hdv%20hd%20rus%20ru]%20[AVC(1280x534(2.4),5 186(5517)kbps,23.976fps)%20R{AC3(0,6,48,448)}%20E{DTS (6,48,1536)}%20SUBS{E}].mkv|9091479113|9C9BE98BCED B8139D0C532DF88B33EB1|/

241

Torrent Links

Torrent Name Torrent Links Watchmen READNFO TS http://www.vertor.com/torrents/910820/Watchmen-READNFO-TS- XviD DEViSE www XviD-DEViSE-NoRar-curren%3B Yestorrent com http://btjunkie.org/torrent/Watchmen-READNFO-TS-XviD-DEViSE- rarbg-com/367860cded0bfd0ae1830151fd0da7e01eb79a3bae35 http://www.torrenthound.com/hash/60cded0bfd0ae1830151fd0da7e01eb7 9a3bae35/torrent-info/Watchmen-2009-TS-TELESYNC-XviD-DEViSE- NoRar%C3%82%C2%A4 http://www.yourbittorrent.com/torrent/5029/Watchmen.READNFO.TS.X viD.DEViSE.html http://www.monova.org/details/2428309/WATCHMEN.2009.TS- TELESYNC.XVID.DEVISE-NORAR%C2%A4%09.html http://www.kickasstorrents.com/t2159689.html http://www.torrentbit.nl/torrent/1660119/Watchmen+READNFO+TS+X viD-DEViSE-%5Brarbg+com%5D/ Watchmen 2009 iTALiAN http://1337x.org/torrent/14393/0/ MD TS XviD www http://www.vertor.com/torrents/720218/Watchmen-2009-iTALiAN-MD- cityofdownloads org TS-XviD-volpebianca-volpebianca http://thepiratebay.org/torrent/4765455/Watchmen.2009.iTALiAN.MD.T S.XviD.volpebianca[volpebianca] http://www.h33t.com/details.php?id=cd5ea57b184c25fa1ff904cdc0b0485 33262e79e http://btjunkie.org/torrent/Watchmen-2009-iTALiAN-MD-TS-XviD- volpebianca- volpebianca/3952cd5ea57b184c25fa1ff904cdc0b048533262e79e http://www.torrenthound.com/hash/cd5ea57b184c25fa1ff904cdc0b04853 3262e79e/torrent-info/Watchmen-2009-iTALiAN-MD-TS-XviD- volpebianca-volpebianca http://www.monova.org/details/2444062/WATCHMEN%202009%20IT ALIAN%20MD%20TS%20XVID%20VOLPEBIANCA%5BVOLPEBI ANCA%5D.html http://torrentportal.com/details/4034645/Watchmen.2009.iTALiAN.MD. TS.XviD.volpebianca.avi.torrent http://www.fulldls.com/torrent-movies-1607237.html http://www.fulldls.com/torrent-movies-1611461.html http://www.kickasstorrents.com/t2177643.html Watchmen TS Xvid Spanish http://www.vertor.com/torrents/681839/Watchman-%28Ts- EspaTorrents com SCREENER%29%28Xvid%29%28Spanish%29%28SpaTorrent- com%29 http://www.btmon.com/Video/Unsorted/Watchman_Ts- SCREENER_Xvid_Spanish_SpaTorrent_com.torrent.html http://www.monova.org/details/2422689/WATCHMAN%20%5BTS- SCREENER%5D%5BXVID%5D%5BSPANISH%5D%5BSPATORRE NT%20COM%5D.html http://extratorrent.com/torrent/1779340/Watchman+Ts- SCREENERXvidSpanishSpa+com.html http://alivetorrents.com/torrent/2354457/watchmen-ts-xvid-spanish- espatorrents- http://www.kickasstorrents.com/t2153183.html Watchmen READNFO TS http://www.newtorrents.info/torrent/5999/Watchmen.READNFO.TS.Xvi XviD DEViSE www D-DEViSE(No.Rars)--cgaurav%E2%84%A2--.html?nopop=1 cubetorrent com http://1337x.org/torrent/14122/0/ http://www.vertor.com/torrents/910815/Watchmen-READNFO-TS-

242

XviD-DEViSE-www-cubetorrent-com http://thepiratebay.org/torrent/4757593/Watchmen_READNFO_TS_Xvi D-DEViSE(No_Rars) http://www.h33t.com/details.php?id=34e77c82515b4cc1c8cfdbdaef0f47d 92c3793e7 http://btjunkie.org/torrent/Watchmen-READNFO-TS-XviD-DEViSE- www-cubetorrent- com/367834e77c82515b4cc1c8cfdbdaef0f47d92c3793e7 http://www.torrenthound.com/hash/34e77c82515b4cc1c8cfdbdaef0f47d9 2c3793e7/torrent-info/Watchmen-READNFO-TS-XviD-DEViSE-No- Rars http://www.yourbittorrent.com/torrent/552402/Watchmen.READNFO.TS .XviD-DEViSE.html http://www.monova.org/details/2428350/WATCHMEN%20READNFO %20TS%20XVID-DEVISE%28NO%20RARS%29--CGAURAV--.html http://www.kickasstorrents.com/t2159792.html http://www.torrentbit.nl/torrent/1669445/Watchmen+READNFO+TS+X viD-DEViSE-%5Bwww+cubetorrent+com%5D/ Watchmen TS XVID STG http://www.newtorrents.info/torrent/5994/Watchmen.TS.XVID.- NoRar curren 1068 2211 .STG(No.Rars)--cgaurav%E2%84%A2--.html?nopop=1 http://1337x.org/torrent/14186/0/ http://www.vertor.com/torrents/694931/Watchmen-TS-XVID-STG- NoRar-curren%3B-1068%2F2211 http://thepiratebay.org/torrent/4759290/Watchmen_TS_XVID_- _STG(No_Rars) http://www.h33t.com/details.php?id=83b00136e6a9d4167c09ed737bf1da ca42e481cb http://www.torrenthound.com/hash/83b00136e6a9d4167c09ed737bf1dac a42e481cb/torrent-info/Watchmen-TS-XVID--STG-No-Rars http://www.yourbittorrent.com/torrent/302141/Watchmen-TS-XVID- STG.html http://www.monova.org/details/2432421/WATCHMEN%20TS%20XVI D%20-%20STG%20NO-RAR.html http://torrentportal.com/details/4028213/Watchmen+TS+XVID- STG.torrent http://alivetorrents.com/torrent/2360899/watchmen-ts-xvid-stg-no-rar http://www.kickasstorrents.com/t2164613.html Watchmen 2009 FRENCH TS http://www.yourbittorrent.com/torrent/303880/Watchmen.2009.FRENCH MD XViD STS MZISYS .TS.MD.XViD-STS.MZISYS.avi.html http://www.monova.org/details/2564544/WATCHMEN%202009%20FR ENCH%20TS%20MD%20XVID-STS%20MZISYS.html http://www.fulldls.com/torrent-movies-1663257.html http://alivetorrents.com/torrent/2468395/watchmen-2009-french-ts-md- xvid-sts-mzisys http://www.kickasstorrents.com/t2446840.html http://www.torrentbit.nl/torrent/614791/Watchmen+2009+FRENCH+TS +MD+XViD-STS+MZISYS/ Watchmen 2009 TS http://www.newtorrents.info/torrent/6007/Watchmen.2009.TS.REencode REencoded XVID Sample Inc d.XVID.(Sample.Inc).ENG-STG.Richmate.html?nopop=1 ENG STG Richmate http://btjunkie.org/torrent/Watchmen-2009-TS-REencoded-XVID-ENG- STG/4594c96c7003055529d1b3263016c4bde1433a4e173e http://www.torrenthound.com/hash/c96c7003055529d1b3263016c4bde14 33a4e173e/torrent-info/Watchmen-2009-TS-REencoded-XVID-ENG- STG http://www.yourbittorrent.com/torrent/145604/Watchmen-2009-TS- REencoded-XVID-ENG-STG.html

243

http://www.monova.org/details/2647688/WATCHMEN%202009%20TS %20REENCODED%20XVID%20%28SAMPLE%20INC%29%20ENG- STG%20RICHMATE.html http://torrentportal.com/details/4162198/Watchmen+2009+TS+REencode d+XVID+ENG-STG.torrent http://www.fulldls.com/torrent-movies-1699305.html http://alivetorrents.com/torrent/2542319/watchmen-2009-ts-reencoded- xvid-sample-inc-eng-stg-richmate http://www.kickasstorrents.com/t2506038.html http://www.torrentbit.nl/torrent/723442/Watchmen+2009+TS+REencode d+XVID+%28Sample+Inc%29+ENG-STG+Richmate/ Watchmen Spanish 2009 TS http://thepiratebay.org/torrent/4757668/Watchmen_(Spanish)_2009_TS- Screener Xvid Mp3 Screener_Xvid-Mp3 http://btjunkie.org/torrent/Watchmen-Spanish-2009-TS-Screener-Xvid- Mp3/1208630117d88341e9262a8190143f0f36e09ed3255f http://www.torrenthound.com/hash/630117d88341e9262a8190143f0f36e 09ed3255f/torrent-info/Watchmen-Spanish-2009-TS-Screener-Xvid-Mp3 http://www.monova.org/details/2428700/WATCHMEN%20%28SPANIS H%29%202009%20TS-SCREENER%20XVID-MP3%09.html http://alivetorrents.com/torrent/5611170/watchmen-spanish-2009-ts- screener-xvid-mp3 http://www.kickasstorrents.com/t2159919.html Watchmen avi 2009 TS http://www.vertor.com/torrents/909027/Watchmen-avi-2009%28TS%29- resynched By resynched-By-DragonLord721%C3%83%C2%82%C3%82%C2%A9- DragonLord721Â© 1774%2F7910 http://thepiratebay.org/torrent/4760763/Watchmen.avi_2009(TS)_resync hed_By_DragonLord721A_ http://www.torrenthound.com/hash/e13bb30eb48f44d43a774164214447f 0d0b7a419/torrent-info/Watchmen-avi-2009-TS-resynched-By- DragonLord721%C3%82%C2%A9 http://www.btmon.com/Video/Movies/Watchmen.avi_2009_TS_resynch ed_By_DragonLord721.torrent.html http://www.monova.org/details/2435525/WATCHMEN.AVI%20%2220 09%22%28TS%29_RESYNCHED%20BY%20DRAGONLORD721.htm l http://www.fulldls.com/torrent-movies-1603191.html http://extratorrent.com/torrent/1786403/Watchmen.avi+%222009%22%2 8TS%29_resynched+By+DragonLord721%A9.html http://alivetorrents.com/torrent/2363326/watchmen.avi- "2009"ts_resynched-by-dragonlord721%C2%A9 http://www.kickasstorrents.com/t2168227.html Watchmen 2009 TS http://www.vertor.com/torrents/692658/Watchmen-2009-TS- TELESYNC XVID STG TELESYNC-XVID-STG-NoRar NoRar http://www.torrenthound.com/hash/79fbc1b49565ac8f01bf1ba29542c8b5 9c3504a0/torrent-info/Watchmen-TS-XVID-STG- NoRar%C3%82%C2%A4 http://www.yourbittorrent.com/torrent/20087/Watchmen-TS-XVID--- STG.html http://www.monova.org/details/2432659/WATCHMEN%20TS%20XVI D%20STG-NORAR.html http://www.fulldls.com/torrent-others-1602071.html http://alivetorrents.com/torrent/2366316/watchmen-2009-ts-telesync- xvid-stg-norar http://www.kickasstorrents.com/t2164808.html Watchmen 2009 Eng TS DivX http://thepiratebay.org/torrent/4760569/Watchmen.[2009.Eng].TS.DivX- LTT LTT

244

http://www.torrenthound.com/hash/a41bfb22a1ccdaca9c2f262882ce56c1 c24886ef/torrent-info/Watchmen-2009-Eng-TS-DivX-LTT http://www.yourbittorrent.com/torrent/112779/Watchmen.%5B2009.Eng %5D.TS.DivX-LTT.html http://www.monova.org/details/2435075/WATCHMEN%20%5B2009%2 0ENG%5D%20TS%20DIVX-LTT.html http://torrentportal.com/details/4031103/Watchmen.%5B2009.Eng%5D. TS.DivX-LTT.torrent http://www.fulldls.com/torrent-movies-1603017.html http://alivetorrents.com/torrent/2363948/watchmen-2009-eng-ts-divx-ltt http://www.kickasstorrents.com/t2168460.html Watchmen TS Screener http://www.monova.org/details/2423552/WATCHMEN%20TS- Spanish Alta Calidad SCREENER%20SPANISH%20ALTA-CALIDAD%09.html http://www.kickasstorrents.com/t2154782.html Watchmen TS Screener 2348 http://www.vertor.com/torrents/910606/Watchmen-TS-Screener- 11150 2348%2F11150 http://www.torrenthound.com/hash/4b14e96003ea45eb6e5b15f3262f4bc 6cb756597/torrent-info/Watchmen-TS-Screener http://www.monova.org/details/2424167/WATCHMEN.TS- SCREENER%09.html http://alivetorrents.com/torrent/5612911/watchmen-ts-screener-- tensiontorrent- http://www.kickasstorrents.com/t2155108.html WATCHMEN 2009 Ts Xvid http://thepiratebay.org/torrent/4873724/WATCHMEN_[2009]_Ts_Xvid_ MultiSubs StB (MultiSubs)_[StB] http://www.torrenthound.com/hash/03852e07a1e22ca2524fe6f10126eb24 3efba284/torrent-info/WATCHMEN-2009-Ts-Xvid-MultiSubs--StB http://www.yourbittorrent.com/torrent/144912/WATCHMEN- %5B2009%5D-Ts-Xvid-%28MultiSubs%29-%5BStB%5D.html http://www.monova.org/details/2644078/WATCHMEN%20%5B2009%5 D%20TS%20XVID%20%28MULTISUBS%29%20%5BSTB%5D%09.h tml http://alivetorrents.com/torrent/7314754/watchmen-2009-ts-xvid- multisubs-stb http://www.torrentbit.nl/torrent/1590111/WATCHMEN+%5B2009%5D+ Ts+Xvid+%28MultiSubs%29+%5BStB%5D/ Watchmen TS ITA STG http://www.vertor.com/torrents/694685/Watchmen-TS-ITA-STG-KVCD- KVCD ResourceRG Reidy ResourceRG-Reidy http://www.torrenthound.com/hash/7a81dc0b93603a39dbacc9e5ced4600 2cff9d971/torrent-info/Watchmen-TS-ITA-STG-KVCD-ResourceRG- Reidy http://www.monova.org/details/2444362/WATCHMEN%20TS%20ITA %20STG%20KVCD%20RESOURCERG%20REIDY%20AVI.html http://alivetorrents.com/torrent/5601840/watchmen-ts-ita-stg-kvcd- resourcerg-reidy http://www.kickasstorrents.com/t2178079.html Watchmen Die Waechter 2009 http://www.torrenthound.com/hash/0b9e9d34c8647b60bc69ede7962439f TS LD German READ NFO 504c86338/torrent-info/Watchmen-Die-Waechter-2009-TS-LD-German- XviD ADLER READ-NFO-XviD-ADLER http://www.btmon.com/Video/Unsorted/Watchmen_Die_Waechter_2009 _TS_LD_German_READ_NFO_XviD-ADLER.torrent.html http://www.monova.org/details/2434422/WATCHMEN%20DIE%20WA ECHTER%202009%20TS%20LD%20GERMAN%20READ%20NFO% 20XVID-ADLER.html http://www.kickasstorrents.com/t2168167.html Watchmen FRENCH TS MD http://www.torrenthound.com/hash/4e2477ff0b6ec984f603e5f4e9df831fb

245

XViD DOLBY WWW 0906310/torrent-info/Watchmen-FRENCH-TS-MD-XViD-DOLBY TORRENTFRANCAIS COM http://www.monova.org/details/2429025/WATCHMEN%20FRENCH%2 0TS%20MD%20XVID- DOLBY%20WWW%20TORRENTFRANCAIS%20COM.html http://alivetorrents.com/torrent/2426001/watchmen-french-ts-md-xvid- dolby--torrentfrancais- http://www.kickasstorrents.com/t2160593.html http://www.torrentbit.nl/torrent/1660683/Watchmen+FRENCH+TS+MD +XViD-DOLBY+WWW+TORRENTFRANCAIS+COM/ Watchmen 2009 TS XViD http://thepiratebay.org/torrent/4758433/Watchmen_(2009)_TS.XViD- CAMERA CAMERA http://btjunkie.org/torrent/Demonoid-Watchmen-2009-TS-XViD- CAMERA/2834db1b126df4c1712dcab03c6f9ef9740dbb302e20 http://www.torrenthound.com/hash/db1b126df4c1712dcab03c6f9ef9740d bb302e20/torrent-info/Watchmen-2009-TS-XViD-CAMERA http://www.btmon.com/Video/Unsorted/Demonoid_Watchmen_2009_TS .XViD-CAMERA.torrent.html http://www.monova.org/details/2430551/%5BDEMONOID%5D%20WA TCHMEN%20%282009%29%20TS.XVID-CAMERA.html http://www.fulldls.com/torrent-movies-1600883.html http://alivetorrents.com/torrent/2359133/demonoid-watchmen-2009- ts.xvid-camera http://www.kickasstorrents.com/t2162193.html Watchmen V O Sub EspaÃ±ol http://thepiratebay.org/torrent/4763821/Watchmen_(V.O.Sub.EspaA_ol)_ 2009 TS Screener Xvid Mp3 2009_TS-Screener_Xvid-Mp3 http://www.torrenthound.com/hash/a1123b493f2033ed74910e788f09e5fc 385397ed/torrent-info/Watchmen-V-O-Sub-Espa%C3%83%C2%B1ol- 2009-TS-Screener-Xvid-Mp3 http://www.yourbittorrent.com/torrent/170923/Watchmen- %28V.O.Sub.Espa%C3%B1ol%29-2009-TS-Screener-Xvid-Mp3- %28Estrenos-Sub.com%29.avi.html http://www.monova.org/details/2441248/WATCHMEN%20%28V.O.SU B.ESPA%C3%B1OL%29%202009%20TS-SCREENER%20XVID- MP3%09.html http://alivetorrents.com/torrent/5603953/watchmen-v.o.sub.espa??ol- 2009-ts-screener-xvid-mp3 http://www.kickasstorrents.com/t2174339.html Watchmen TS XVID STG http://www.newtorrents.info/torrent/5995/Watchmen.TS.XVID.-.STG-- www seedmore org cgaurav%E2%84%A2--.html?nopop=1 http://1337x.org/torrent/14181/0/ http://thepiratebay.org/torrent/4758968/Watchmen_TS_XVID_-_STG http://www.torrenthound.com/hash/24e93033fec0fc5380d62ec4bbb3e547 f2b5ffc6/torrent-info/Watchmen-TS-XVID--STG http://www.btmon.com/Video/Unsorted/Watchmen_TS_XVID_- _STG_www_seedmore_org.torrent.html http://www.yourbittorrent.com/torrent/553434/Watchmen-TS-XVID--- STG.html http://www.monova.org/details/2431827/WATCHMEN%20TS%20XVI D%20-%20STG%5BWWW%20SEEDMORE%20ORG%5D.html http://alivetorrents.com/torrent/2360378/watchmen-ts-xvid-stg-data- force. http://www.kickasstorrents.com/t2164026.html Watchmen TS XVID STG NO http://www.vertor.com/torrents/1213463/Watchmen-TS-XVID-STG- RARS %28-NO-RARS-%29 http://thepiratebay.org/torrent/4767464/Watchmen_TS_XVID_- _STG_(_NO_RARS_)

246

http://www.torrenthound.com/hash/d05c26b7afc391fed5e952c7d93f42a5 90e77874/torrent-info/Watchmen-TS-XVID--STG--NO-RARS http://www.monova.org/details/2447831/WATCHMEN%20TS%20XVI D%20-%20STG%20%28%20NO%20RARS%20%29.html http://torrentportal.com/details/4037134/Watchmen+TS+XVID+- +STG.torrent http://alivetorrents.com/torrent/2372610/watchmen-ts-xvid-stg--no-rars- http://www.kickasstorrents.com/t2181304.html Watchmen TS Xvid FOX Watchmen Die Waechter 2009 http://thepiratebay.org/torrent/4756151/Watchmen.Die.Waechter.2009.T TS LD German XviD TRG S.LD.German.XviD-TRG http://www.torrenthound.com/hash/472197920b3ebde854d15b223d26bef 5172b3af8/torrent-info/Watchmen-Die-Waechter-2009-TS-LD-German- XviD-TRG http://www.btmon.com/Video/Movies/Watchmen.Die.Waechter.2009.TS .LD.German.XviD-TRG617299654236.981.torrent.html http://www.monova.org/details/2424279/WATCHMEN.DIE.WAECHTE R.2009.TS.LD.GERMAN.XVID-TRG%09.html http://alivetorrents.com/torrent/5612770/watchmen.die.waechter.2009.ts.l d.german.xvid-trg http://www.kickasstorrents.com/t2155806.html Watchmen 2009 TS NL subs http://thepiratebay.org/torrent/4774542/Watchmen_(2009)_TS_NL_subs NLT Release _-_NLT_Release_- http://www.torrenthound.com/hash/6d36056304171aafd1040009cfab1cea 07482370/torrent-info/Watchmen-2009-TS-NL-subs--NLT-Release http://www.btmon.com/Video/Movies/Watchmen_2009_TS_NL_subs_- _NLT_Release_-.torrent.html http://www.monova.org/details/2460549/WATCHMEN%20%282009%2 9%20TS%20NL%20SUBS%20-%20NLT%20RELEASE%20-.html http://torrentportal.com/details/4044942/Watchmen+-+DviX+- +NLT+RELEASE+-.torrent http://alivetorrents.com/torrent/2382635/watchmen-2009-ts-nl-subs-nlt- release-- http://www.kickasstorrents.com/t2194612.html Watchmen TS For iPhone http://www.torrenthound.com/hash/4c29d0396b5af7d25ceb4edcc4aa399f iPython mp4 bee58e9d/torrent-info/Watchmen-TS-For-iPhone--iPython-mp4 http://www.monova.org/details/2459715/WATCHMEN%20%5BTS%5D %5BFOR%20IPHONE%5D-%5BIPYTHON%5D%20MP4.html http://alivetorrents.com/torrent/2381810/watchmen-tsfor-iphone-ipython- mp4 http://www.kickasstorrents.com/t2450974.html Watchmen TS XviD W A L http://thepiratebay.org/torrent/4791171/Watchmen.TS.XviD-W.A.L http://www.torrenthound.com/hash/ac95a991830e1a4b3d02b318d4c17a7 f37c55a9d/torrent-info/Watchmen-TS-XviD-W-A-L http://www.monova.org/details/2488777/WATCHMEN.TS.XVID- W.A.L%09.html http://torrentportal.com/details/4063782/Watchmen.TS.XviD- W.A.L.torrent http://www.fulldls.com/torrent-movies-1627972.html http://alivetorrents.com/torrent/2406141/watchmen-ts-xvid-w-a-l http://www.kickasstorrents.com/t2231374.html Watchmen 2009 TS Fixed http://www.torrenthound.com/hash/54e52bb0cdc714514f0655796332998 DivX www leuboo com 415a0b6ff/torrent-info/Watchmen-2009-TS-Fixed-DivX http://www.monova.org/details/2466655/WATCHMEN%202009%20TS %20FIXED%20DIVX-%5BWWW%20LEUBOO%20COM%5D.html http://alivetorrents.com/torrent/2388352/watchmen-2009-ts-fixed-divx--

247

leuboo- http://www.kickasstorrents.com/t2202687.html Watchmen READNFO TS http://www.h33t.com/details.php?id=52e83b299a8964024381c8fc9a6fcd XviD DEViSE 7754698b87 http://www.torrenthound.com/hash/52e83b299a8964024381c8fc9a6fcd7 754698b87/torrent-info/Watchmen-READNFO-TS-XviD-DEViSE http://www.btmon.com/Video/Unsorted/Watchmen_READNFO_TS_Xvi D_DEViSE.torrent.html http://www.monova.org/details/2930985/WATCHMEN%20READNFO %20TS%20XVID%20DEVISE.html http://www.fulldls.com/torrent-movies-1832431.html http://www.kickasstorrents.com/t2831401.html Watchmen TS Screener HQ http://thepiratebay.org/torrent/4755714/Watchmen.TS-Screener- Xvid Mp3 Spanish HQ.Xvid-Mp3.Spanish. http://www.torrenthound.com/hash/ab97076de15a0be4a673bfe6e992d70 d92b0bca3/torrent-info/Watchmen-TS-Screener-HQ-Xvid-Mp3-Spanish http://www.monova.org/details/2423589/WATCHMEN.TS- SCREENER-HQ.XVID-MP3.SPANISH.%09.html http://alivetorrents.com/torrent/5613393/watchmen.ts-screener-hq.xvid- mp3.spanish http://www.kickasstorrents.com/t2154943.html Watchmen TS Spanish 2009 http://www.vertor.com/torrents/1213840/Watchmen-TS-SCREENER- www neodk com Spanish-2009-www-neodk-com http://www.monova.org/details/2426885/WATCHMEN%20%20%5BTS- SCREENER%5D%5BSPANISH%5D%5B2009%5D%5BWWW%20NE ODK%20COM%5D.html http://www.fulldls.com/torrent-movies-1599281.html http://alivetorrents.com/torrent/2356257/watchmen--ts-spanish2009- http://www.kickasstorrents.com/t2451835.html Watchmen TS Xvid Camera http://www.monova.org/details/2429839/WATCHMEN-TS-XVID- 2cds ESPATORRENTS com CAMERA-2CDS-%5BESPATORRENTS%20COM%5D.html http://alivetorrents.com/torrent/2358386/watchmen-ts-xvid-camera-2cds- espatorrents- http://www.kickasstorrents.com/t2161378.html Watchmen TS XViD CAMERA Watchmen TS Xvid FOX NO http://www.vertor.com/torrents/693239/Watchmen-TS-Xvid-FOX-%28- RARS NO-RARS-%29 http://thepiratebay.org/torrent/4764336/Watchmen.TS.Xvid- FOX_(_NO_RARS_) http://www.torrenthound.com/hash/27760f9711bb780ec9420663387efcc 0a7cfc70a/torrent-info/Watchmen-TS-Xvid-FOX--NO-RARS http://www.monova.org/details/2442060/WATCHMEN.TS.XVID- FOX%20%28%20NO%20RARS%20%29%09.html http://alivetorrents.com/torrent/2368067/watchmen.ts.xvid-fox--no-rars- http://www.kickasstorrents.com/t2175285.html Watchmen TS Xvid FOX http://www.monova.org/details/2442714/WATCHMEN%20TS%20XVI www supreme speeds com D-FOX%20WWW%20SUPREME-SPEEDS%20COM.html http://torrentportal.com/details/4033567/Watchmen.TS.Xvid-FOX.torrent http://alivetorrents.com/torrent/2368514/watchmen-ts-xvid-fox-- http://www.kickasstorrents.com/t2176049.html Watchmen Die Waechter 2009 http://www.torrenthound.com/hash/0921c03fdd379b2f5479e3244e29a9d TS LD German XviD TRG 9b24c521f/torrent-info/Watchmen-Die-Waechter-2009-TS-LD-German- www supreme speeds com 2 XviD--TRG http://www.monova.org/details/2422685/WATCHMEN%20DIE%20WA

248

ECHTER%202009%20TS%20LD%20GERMAN%20XVID- TRG%20WWW%20SUPREME-SPEEDS%20COM%282%29.html http://alivetorrents.com/torrent/2352831/watchmen-die-waechter-2009- ts-ld-german-xvid-trg--2 http://www.kickasstorrents.com/t2153109.html Watchmen FRENCH 2009 http://www.torrenthound.com/hash/8f02ec597d642209749fe005b7a8567 screener cam TS MD e7555242e/torrent-info/Watchmen-FRENCH-2009-screener-cam-TS- REPACK 1CD XViD STS MD-REPACK-1CD-XViD-STS-avi www miragetorrent com avi http://www.monova.org/details/2424914/WATCHMEN%20FRENCH%2 0TS%20MD%20REPACK%201CD%20XVID- STS%20%28%20%20WWW.MIRAGETORRENT.COM%20%20%29% 20AVI.html http://alivetorrents.com/torrent/2519559/watchmen-french-2009- screener-cam-ts-md-repack-1cd-xvid-sts.miragetorrent.-avi http://www.kickasstorrents.com/t3005616.html Watchmen READNFO TS http://www.newtorrents.info/torrent/6000/Watchmen.READNFO.TS.Xvi XviD DEViSE WWW D-DEViSE--cgaurav%E2%84%A2--.html?nopop=1 SCENEPIRATES COM http://1337x.org/torrent/14121/0/ http://thepiratebay.org/torrent/4757547/Watchmen_READNFO_TS_Xvi D-DEViSE http://www.btmon.com/Video/Movies/Watchmen_READNFO_TS_XviD -DEViSE.torrent.html http://www.yourbittorrent.com/torrent/553725/Watchmen.READNFO.TS .XviD-DEViSE.html http://www.monova.org/details/2428148/WATCHMEN%20READNFO %20TS%20XVID-DEVISE--CGAURAV--.html http://extratorrent.com/torrent/1782802/Watchmen+READNFO+TS+Xvi D-DEViSE%28.html http://www.kickasstorrents.com/t2159651.html Watchmen READNFO TS http://www.btmon.com/Video/Unsorted/Watchmen_READNFO_TS_Xvi XviD DEViSE D-DEViSE499612161782.669.torrent.html http://www.monova.org/details/2428939/WATCHMEN%20READNFO %20TS%20XVID-DEVISE.html http://www.kickasstorrents.com/t2160451.html Watchmen 2009 TS DEViSE http://alivetorrents.com/torrent/2358154/watchmen-2009-ts-devise- Kingdom KvCD by Bexley kingdom-kvcd-by-bexley http://www.kickasstorrents.com/t2160848.html Watchmen avi quot 2009 quot http://www.vertor.com/torrents/1213655/Watchmen-avi-quot%3B2009- TS New Audio resynched By quot%3B%28TS%29-New-Audio-resynched-By-DragonLord721- DragonLord721 copy copy%3B http://www.torrenthound.com/hash/2e716d061c28e48f9f0dc9219bbfe08c 22d3fb22/torrent-info/Watchmen-avi-2009-TS-New-Audio-resynched- By-DragonLord7 http://www.btmon.com/Video/Movies/Watchmen.avi_2009_TS_New_A udio_resynched_By_DragonLord7.torrent.html http://www.monova.org/details/2440190/WATCHMEN.AVI%20%2220 09%22%28TS%29_NEW%20AUDIO_RESYNCHED%20BY%20DRA GONLORD721.html http://alivetorrents.com/torrent/2366804/watchmen.avi-"2009"ts_new- audio_resynched-by-dragonlord721%C2%A9 http://www.kickasstorrents.com/t2173439.html Watchmen 2009 TS 1088kbps http://www.torrenthound.com/hash/a078664abe0dafc093d1d82ec3eda2ff PMTORRENTS COM 41caf904/torrent-info/Watchmen-2009-TS-1088kbps-PMTORRENTS- IPHONE IPOD ITOUCH COM-IPHONE-IPOD-ITOUCH-ZUNE-PSP ZUNE PSP http://www.yourbittorrent.com/torrent/941366/Watchmen.2009.TS.1088k bps.%28PMTORRENTS.COM%29.IPHONE.IPOD.ITOUCH.ZUNE.PS

249

P.html http://www.monova.org/details/2447048/WATCHMEN%202009%20TS %201088KBPS%20%20PMTORRENTS%20COM%20%20IPHONE%2 0IPOD%20ITOUCH%20ZUNE%20PSP.html http://torrentportal.com/details/4036938/Watchmen.2009.TS.1088kbps.% 28PMTORRENTS.COM%29.IPHONE.IPOD.ITOUCH.ZUNE.PSP.torre nt http://alivetorrents.com/torrent/2371776/watchmen-2009-ts-1088kbps-- pmtorrents---iphone-ipod-itouch-zune-psp http://www.kickasstorrents.com/t2180557.html http://www.torrents.net/torrent/213477/Watchmen.2009.TS.1088kbps.%2 8PMTORRENTS.COM%29.IPHONE.IPOD.ITOUCH.ZUNE.PSP/ Watchmen 2009 TS XViD http://www.torrenthound.com/hash/7ec533f718a3899fe4939086ffcf33b1 iPython avi 6edba0af/torrent-info/Watchmen-2009-TS-XViD-iPython-avi http://www.monova.org/details/2459799/WATCHMEN%202009%20%5 BTS%5D%5BXVID%5D%5BIPYTHON%5D%20AVI.html http://alivetorrents.com/torrent/2381722/watchmen-2009-tsxvidipython- avi http://www.kickasstorrents.com/t2215962.html http://www.torrentbit.nl/torrent/289114/Watchmen+2009+TSXViDiPyth on+avi/ Watchmen TS Screener http://thepiratebay.org/torrent/4757727/Watchmen[TS- Spanish 2009 Screener][Spanish][2009] http://www.torrenthound.com/hash/4b237faff1b24398cc9eba644a57cdbf 1978359a/torrent-info/Watchmen-TS-Screener-Spanish-2009 http://www.monova.org/details/2428737/WATCHMEN%5BTS- SCREENER%5D%5BSPANISH%5D%5B2009%5D%09.html http://alivetorrents.com/torrent/5610616/watchments- screenerspanish2009 http://www.kickasstorrents.com/t2160092.html Watchmen TS XViD http://www.vertor.com/torrents/1213599/Watchmen-TS-XViD- CAMERA www torrent loco CAMERA-torrent-loco-com-ar-154%2F632 com ar http://www.torrenthound.com/hash/90117b317f537c765186e7299e13164 3cbf004d5/torrent-info/Watchmen-TS-XViD-CAMERA-ar http://www.yourbittorrent.com/torrent/766558/Watchmen.TS.XViD- CAMERA.www.torrent-loco.com.ar.html http://www.monova.org/details/2429997/WATCHMEN%20TS%20XVI D-CAMERA%20WWW%20TORRENT-LOCO%20COM%20AR.html http://torrentportal.com/details/4024809/Watchmen.TS.XViD- CAMERA.www.torrent-loco.com.ar.torrent http://alivetorrents.com/torrent/2358586/watchmen-ts-xvid-camera---ar http://www.kickasstorrents.com/t2161594.html Watchmen TS XVID STG http://thepiratebay.org/torrent/4763250/Watchmen_TS_XVID_-_STG www torrentsforall net http://www.torrenthound.com/hash/d4b3730f6dfdc92f56fa29538f21b1fa 829fe06d/torrent-info/Watchmen-TS-XVID--STG http://www.btmon.com/Video/Unsorted/Watchmen_TS_XVID_- _STG_Strikeys-torrents.com_Staff_required.torrent.html http://www.monova.org/details/2432898/WATCHMEN%20TS%20XVI D%20-%20STG%20%7BSTRIKEYS- TORRENTS.COM%7D%20STAFF%20REQUIRED.html http://torrentportal.com/details/4033184/Watchmen+TS+XVID+- +STG.torrent http://www.fulldls.com/torrent-movies-1603354.html http://extratorrent.com/torrent/1785103/Watchmen+TS+XVID+- +STG+%7BStrikeys+s.com%7D+Staff+required.html http://alivetorrents.com/torrent/2363625/watchmen-ts-xvid-stg-

250

http://www.kickasstorrents.com/t2164983.html Watchmen TS XVID STG http://www.btmon.com/Video/Unsorted/Watchmen_TS_XVID_- www pretome net _STG_www_pretome_net.torrent.html http://www.monova.org/details/2434157/WATCHMEN%20TS%20XVI D%20-%20STG%5BWWW%20PRETOME%20NET%5D.html http://www.fulldls.com/torrent-movies-1602262.html http://alivetorrents.com/torrent/2361697/watchmen-ts-xvid-stg- http://www.kickasstorrents.com/t2165891.html watchmen TS Xvid JamBo http://www.torrenthound.com/hash/19ee89813c1e498ef7de2ccb6f6dda49 mazysnuthouse org 3058b12c/torrent-info/watchmen-TS-Xvid--JamBo-mazysnuthouse-org http://www.btmon.com/Video/Unsorted/watchmen_TS_Xvid_-_JamBo- mazysnuthouse_org.torrent.html http://www.monova.org/details/2435328/WATCHMEN%20TS%20XVI D%20-%20JAMBO-MAZYSNUTHOUSE%20ORG.html http://alivetorrents.com/torrent/2363086/watchmen-ts-xvid-jambo- mazysnuthouse- http://www.kickasstorrents.com/t2167966.html watchmen TS Xvid JamBo http://www.torrenthound.com/hash/72a60b165073243c887ab4ac76b812f www supreme speeds com 228b2e2d3/torrent-info/watchmen-TS-Xvid--JamBo http://www.monova.org/details/2435299/WATCHMEN%20TS%20XVI D%20-%20JAMBO%20%20WWW%20SUPREME- SPEEDS%20COM.html http://alivetorrents.com/torrent/2363064/watchmen-ts-xvid-jambo- http://www.kickasstorrents.com/t2167929.html Watchmen TS FullMovie STG http://1337x.org/torrent/14259/0/ KVCD ResourceRG Reidy http://thepiratebay.org/torrent/4760881/Watchmen_TS_FullMovie_STG_ KVCD_ResourceRG_Reidy http://www.h33t.com/details.php?id=8354cf075f68a6e0926ee9634f60ed d3b55a6d55 http://btjunkie.org/torrent/Watchmen-TS-FullMovie-STG-KVCD- ResourceRG-Reidy/39528354cf075f68a6e0926ee9634f60edd3b55a6d55 http://www.torrenthound.com/hash/8354cf075f68a6e0926ee9634f60edd3 b55a6d55/torrent-info/Watchmen-TS-FullMovie-STG-KVCD- ResourceRG-Reidy http://www.yourbittorrent.com/torrent/271799/Watchmen-TS-FullMovie- STG-KVCD-ResourceRG-Reidy.html http://www.monova.org/details/2435772/WATCHMEN%20TS%20FUL LMOVIE%20STG%20KVCD%20RESOURCERG%20REIDY%09.html http://extratorrent.com/torrent/1786983/Watchmen+TS+FullMovie+STG +KVCD+ResourceRG+Reidy.html http://alivetorrents.com/torrent/2363637/watchmen-ts-fullmovie-stg- kvcd-resourcerg-reidy http://www.kickasstorrents.com/t2168525.html watchmen TS Xvid JamBo http://www.monova.org/details/2435857/WATCHMEN%20TS%20XVI mazysmadhouse net D%20-%20JAMBO%20MAZYSMADHOUSE%20NET.html http://www.fulldls.com/torrent-movies-1603356.html http://www.kickasstorrents.com/t2168601.html Watchmen TS 2009 RMVB http://thepiratebay.org/torrent/4763971/Watchmen_[TS][2009][RMVB][ ksalman ksalman] http://www.torrenthound.com/hash/05168c2966c104eb4c4026810c315f1 fb4865974/torrent-info/Watchmen-TS-2009-RMVB-ksalman http://www.monova.org/details/2441512/WATCHMEN%20%5BTS%5D %5B2009%5D%5BRMVB%5D%5BKSALMAN%5D%09.html http://alivetorrents.com/torrent/5603557/watchmen-ts2009rmvbksalman http://www.kickasstorrents.com/t2174605.html

251

Watchmen TS XViD http://thepiratebay.org/torrent/4758727/Watchmen.TS.XViD-CAMERA CAMERA http://www.torrenthound.com/hash/3966f01dc21bb234618e8fa522801f4 3a2d9a011/torrent-info/Watchmen-TS-XViD-CAMERA-FilmsBT-com http://www.monova.org/details/2431422/WATCHMEN.TS.XVID- CAMERA.%5BFILMSBT.COM%5D%09.html http://torrentportal.com/details/4027867/Watchmen.TS.XViD- CAMERA.%5BFilmsBT.com%5D.torrent http://alivetorrents.com/torrent/5608860/watchmen.ts.xvid-camera http://www.kickasstorrents.com/t2163462.html Watchmen TS Xvid FOX http://www.monova.org/details/2442362/WATCHMEN%20TS%20XVI teamfx org D-FOX-%5BTEAMFX%20ORG%5D.html http://alivetorrents.com/torrent/2368226/watchmen-ts-xvid-fox-teamfx- http://www.kickasstorrents.com/t2175555.html Watchmen TS XVID STG http://thepiratebay.org/torrent/4777705/Watchmen_TS_XVID_- torrent _STG.torrent http://www.torrenthound.com/hash/abe2bfde210584379f5d161e877cc8e0 47fab23d/torrent-info/Watchmen-TS-XVID--STG http://www.monova.org/details/2465548/WATCHMEN_TS_XVID_- _STG%09.html http://torrentportal.com/details/4048891/Watchmen+TS+XVID+- +STG.torrent http://www.fulldls.com/torrent-others-1618182.html http://alivetorrents.com/torrent/5591166/watchmen_ts_xvid_-_stg http://www.kickasstorrents.com/t2201463.html watchmen TS Xvid JamBo http://www.monova.org/details/2521157/WATCHMEN%20TS%20XVI www darknesstorrents com D%20-%20JAMBO-WWW.DARKNESSTORRENTS.COM.html http://alivetorrents.com/torrent/2432160/watchmen-ts-xvid-jambo- .darknesstorrents. http://www.kickasstorrents.com/t2448105.html Watchmen Die Waechter TS http://www.monova.org/details/2420496/WATCHMEN%20DIE%20WA LD German XviD TRG ECHTER%20TS%20LD%20GERMAN%20XVID-TRG.html http://alivetorrents.com/torrent/2351873/watchmen-die-waechter-ts-ld- german-xvid-trg http://www.kickasstorrents.com/t2151517.html Watchmen Die Waechter TS http://www.btmon.com/Video/Unsorted/Watchmen_Die_Waechter_TS_ LD German XviD TRG LD_German_XviD-TRG31749601883.494.torrent.html http://www.monova.org/details/2420544/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN%20XVID-TRG.html http://alivetorrents.com/torrent/2351904/watchmen-die-waechter-ts-ld- german-xvid-trg http://www.kickasstorrents.com/t2152572.html Watchmen Die Waechter TS http://www.torrenthound.com/hash/7cb5cddaae58b8523585af0a5ed0f6b7 LD German XviD TRG torrent e8902fae/torrent-info/Watchmen-Die-Waechter-TS-LD-German-XviD-- mnvv2 info TRG http://www.monova.org/details/2422479/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN%20XVID- TRG%20TORRENT%20%5BMNVV2%20INFO%5D.html http://alivetorrents.com/torrent/2352677/watchmen-die-waechter-ts-ld- german-xvid-trg-torrent-mnvv2- http://www.kickasstorrents.com/t2152872.html Watchmen Die Waechter TS http://www.torrenthound.com/hash/4bbb4f0c98a7c57005ed971410ebb62 LD German XviD TRG 855973f29/torrent-info/Watchmen-Die-Waechter-TS-LD-German-XviD- -TRG http://www.btmon.com/Video/Unsorted/Watchmen_Die_Waechter_TS_ LD_German_XviD-TRG603133807179.454.torrent.html

252

http://www.monova.org/details/2423095/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN%20XVID-TRG.html http://alivetorrents.com/torrent/2353306/watchmen-die-waechter-ts-ld- german-xvid-trg http://www.kickasstorrents.com/t2155226.html www torrent god com http://www.btmon.com/Video/Unsorted/Watchmen_READNFO_TS_Xvi Watchmen READNFO TS D-DEViSE_Data-Force.net.torrent.html XviD DEViSE http://www.monova.org/details/2428210/WATCHMEN%20READNFO %20TS%20XVID-DEVISE%20DATA-FORCE.NET.html http://www.kickasstorrents.com/t2159706.html Watchmen READNFO TS http://www.monova.org/details/2428639/WATCHMEN%20READNFO XviD DEViSE www PeerDen %20TS%20XVID- com DEVISE%20%5BWWW%20PEERDEN%20COM%5D.html http://www.kickasstorrents.com/t2451685.html Watchmen READNFO TS http://www.monova.org/details/2428245/WATCHMEN%20READNFO XviD DEViSE www supreme %20TS%20XVID-DEVISE%20WWW%20SUPREME- speeds com SPEEDS%20COM.html http://www.kickasstorrents.com/t2451788.html Watchmen Die Waechter TS http://www.torrenthound.com/hash/72492f0a7a993c81c303176e00324dd LD German READ NFO e8d0b29be/torrent-info/Watchmen-Die-Waechter-TS-LD-German- REPACK XviD ADLER READ-NFO-REPACK-XviD-ADLER http://www.monova.org/details/2437296/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN%20READ%20NFO%20REPA CK%20XVID-ADLER.html http://www.kickasstorrents.com/t2170263.html Watchmen READNFO TS http://thepiratebay.org/torrent/4758130/Watchmen_READNFO_TS_Kvc Kvcd By Flames d_By_Flames http://www.monova.org/details/2429939/WATCHMEN%20READNFO %20TS%20KVCD%20BY%20FLAMES%09.html http://www.kickasstorrents.com/t2161285.html Watchmen READNFO TS http://thepiratebay.org/torrent/4758158/Watchmen.READNFO.TS.XviD- XviD DEViSE DEViSE. http://www.torrenthound.com/hash/7c0180d9ba8a3bafd39332658be449d c1f64025e/torrent-info/Watchmen-READNFO-TS-XviD-DEViSE- FilmsBT-com http://www.monova.org/details/2429829/WATCHMEN%20READNFO %20TS%20XVID-DEVISE%20%5BFILMSBT%20COM%5D.html http://www.kickasstorrents.com/t2161376.html Watchmen Die Waechter TS http://www.torrenthound.com/hash/971c8eaa7829ccb3d8c315f68f965101 LD German XviD TRG 91ea1dcb/torrent-info/Watchmen-Die-Waechter-TS-LD-German-XviD- TRG http://www.monova.org/details/2608803/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN%20XVID-TRG.html http://alivetorrents.com/torrent/2506798/watchmen-die-waechter-ts-ld- german-xvid-trg http://www.kickasstorrents.com/t2470519.html Watchmen Dvd r SecretArea http://www.monova.org/details/3139821/WATCHMEN-DVD-R- CREW Area51 ts info SECRETAREA-CREW%5BAREA51-TS%20INFO%5D.html http://alivetorrents.com/torrent/2975258/watchmen-dvd-r-secretarea- crewarea51-ts- http://www.kickasstorrents.com/t3033237.html Watchmen TS XviD DEViSE http://www.monova.org/details/2428600/WATCHMEN%20TS%20XVI movie shack org D%20DEVISE%20HTTP%3A%2F%2FMOVIE-SHACK.ORG.html http://alivetorrents.com/torrent/2357687/watchmen-ts-xvid-devise http://www.kickasstorrents.com/t2160069.html

253

Watchmen TS XViD http://www.vertor.com/torrents/1984261/Watchmen-TS-XViD- CAMERA scenepirates com CAMERA%28scenepirates-com%29 http://www.monova.org/details/2430440/WATCHMEN%20TS%20XVI D-CAMERA%28SCENEPIRATES.COM%29.html http://alivetorrents.com/torrent/2359094/watchmen-ts-xvid- camerascenepirates. http://www.kickasstorrents.com/t2162054.html Watchmen TS XViD http://www.torrenthound.com/hash/e731530d4ae5f0ca839d4ac65328fd8d CAMERA mazysnuthouse org 3f566f06/torrent-info/Watchmen-TS-XViD-CAMERA-mazysnuthouse- org http://www.monova.org/details/2429902/WATCHMEN%20TS%20XVI D-CAMERA%20MAZYSNUTHOUSE%20ORG.html http://alivetorrents.com/torrent/2358414/watchmen-ts-xvid-camera- mazysnuthouse- http://www.kickasstorrents.com/t2161430.html Watchmen TS XViD http://www.torrenthound.com/hash/b53ff79047f4bff7f8953612b911e73e CAMERA www supreme 95da2089/torrent-info/Watchmen-TS-XViD--CAMERA speeds com http://www.monova.org/details/2429909/WATCHMEN%20TS%20XVI D-CAMERA%20WWW%20SUPREME-SPEEDS%20COM.html http://alivetorrents.com/torrent/2358404/watchmen-ts-xvid-camera-- http://www.kickasstorrents.com/t2161807.html Watchmen TS XViD http://www.monova.org/details/2430979/WATCHMEN%20TS%20XVI CAMERA,mazysmadhouse D-CAMERA%2CMAZYSMADHOUSE%20NET.html net http://www.kickasstorrents.com/t2162894.html Watchmen TS XViD http://www.torrenthound.com/hash/615ed5a13368fb139531df7a9e27548 CAMERA www pretome net 112bd4d2b/torrent-info/Watchmen-TS-XViD--CAMERA http://www.monova.org/details/2431130/WATCHMEN%20TS%20XVI D-CAMERA%5BWWW%20PRETOME%20NET%5D.html http://www.fulldls.com/torrent-movies-1601123.html http://alivetorrents.com/torrent/2359670/watchmen-ts-xvid-camera- http://www.kickasstorrents.com/t2162992.html Watchmen TS XViD http://www.monova.org/details/2431824/WATCHMEN%20TS%20XVI CAMERA www heavens D-CAMERA%5BWWW%20HEAVENS-ABOVE%20BIZ%5D.html above biz http://alivetorrents.com/torrent/2360228/watchmen-ts-xvid-camera-biz http://www.kickasstorrents.com/t2163856.html Watchmen TS XVID STG http://www.monova.org/details/2431869/WATCHMEN%20TS%20XVI teamfx org D%20-%20STG-%5BTEAMFX%20ORG%5D.html http://alivetorrents.com/torrent/2360355/watchmen-ts-xvid-stg-teamfx- http://www.kickasstorrents.com/t2163962.html Watchmen TS XVID STG http://www.monova.org/details/2431855/WATCHMEN%20TS%20XVI XTREMESPEEDs NET D%20-%20STG%20XTREMESPEEDS.NET.html http://alivetorrents.com/torrent/2360351/watchmen-ts-xvid-stg- xtremespeeds. http://www.kickasstorrents.com/t2163960.html Watchmen TS XVID STG http://www.monova.org/details/2431870/WATCHMEN%20TS%20XVI www supreme speeds com D%20-%20STG%20%20WWW.SUPREME-SPEEDS%20COM.html http://alivetorrents.com/torrent/2360383/watchmen-ts-xvid-stg- http://www.kickasstorrents.com/t2164030.html Watchmen TS XVID STG http://www.monova.org/details/2432011/WATCHMEN%20TS%20XVI mazysnuthouse org D%20-%20STG%20MAZYSNUTHOUSE%20ORG.html http://alivetorrents.com/torrent/2360445/watchmen-ts-xvid-stg- mazysnuthouse- http://www.kickasstorrents.com/t2164269.html Watchmen TS XVID STG http://www.monova.org/details/2432287/WATCHMEN%20TS%20XVI D%20-%20STG.html

254

http://alivetorrents.com/torrent/2360739/watchmen-ts-xvid-stg http://www.kickasstorrents.com/t2164448.html Watchmen TS XViD http://www.monova.org/details/2431761/WATCHMEN%20TS%20XVI CAMERA Torrentzilla org D-CAMERA%20%5B%20TORRENTZILLA.ORG%20%5D.html http://www.kickasstorrents.com/t2163941.html Watchmen TS XViD http://www.monova.org/details/2431488/WATCHMEN%20TS%20XVI CAMERA teamfx org D-CAMERA-%5BTEAMFX%20ORG%5D.html http://alivetorrents.com/torrent/2359991/watchmen-ts-xvid-camera- teamfx- http://www.kickasstorrents.com/t2161617.html Watchmen TS XVID STG http://www.monova.org/details/2433265/WATCHMEN%20TS%20XVI www sparksden org D%20-%20STG%20%5BWWW.SPARKSDEN.ORG%5D.html http://www.fulldls.com/torrent-movies-1602098.html http://alivetorrents.com/torrent/2361470/watchmen-ts-xvid-stg- http://www.kickasstorrents.com/t2165569.html Watchmen TS/CAM http://thepiratebay.org/torrent/4759666/Watchmen_TS_CAM http://www.torrenthound.com/hash/207c971eeef26b0a1ee3f1bf52786011 0ea7e53e/torrent-info/Watchmen-TS-CAM http://www.monova.org/details/2434487/WATCHMEN%20TS%2FCAM %09.html http://alivetorrents.com/torrent/5608063/watchmen-ts-cam http://www.kickasstorrents.com/t2165734.html Watchmen TS XVID STG http://www.torrenthound.com/hash/5c992cd8e2511fb54e7801620649f82 ceasers palace info 7eacdfe34/torrent-info/Watchmen-TS-XVID--STG-ceasers-palace-info http://www.monova.org/details/2434129/WATCHMEN%20TS%20XVI D%20-%20STG%20CEASERS-PALACE%20INFO.html http://torrentportal.com/details/4028170/Watchmen+TS+XVID+- +STG.torrent http://alivetorrents.com/torrent/2361662/watchmen-ts-xvid-stg-ceasers- palace- http://www.kickasstorrents.com/t2165829.html Watchmen TS STG http://www.torrenthound.com/hash/9ac9ae4bc7a486edc38b06f7ebc640f7 8e3e7b3e/torrent-info/Watchmen-TS-STG http://www.monova.org/details/2439484/WATCHMEN%20TS- STG.html http://torrentportal.com/details/4031890/Watchmen+TS+XVID+- +STG.torrent http://alivetorrents.com/torrent/2366271/watchmen-ts-stg http://www.kickasstorrents.com/t2172787.html Watchmen TS Xvid FOX http://www.torrenthound.com/hash/c5ffc86f9ed17a7c2b3b735f24467e8f 5728429f/torrent-info/Watchmen-TS-Xvid-FOX http://www.monova.org/details/2443070/WATCHMEN%20TS%20XVI D-FOX.html http://alivetorrents.com/torrent/2368718/watchmen-ts-xvid-fox http://www.kickasstorrents.com/t2176414.html Watchmen TS XViD http://www.torrenthound.com/hash/dbd11880a87eb47b4e1aab73887b5dd CAMERA www torrentsforall f4af547da/torrent-info/Watchmen-TS-XViD-CAMERA net http://www.monova.org/details/2454541/WATCHMEN%20TS%20XVI D-CAMERA%20%5BWWW.TORRENTSFORALL.NET%5D.html http://alivetorrents.com/torrent/2377639/watchmen-ts-xvid-camera- .torrentsforall. http://www.kickasstorrents.com/t2188374.html Watchmen TS XVID STG http://www.torrenthound.com/hash/4b60d9883377a83dee64a00e3505b1e pirates cove biz b0754ac87/torrent-info/Watchmen-TS-XVID--STG-pirates-cove-biz http://www.monova.org/details/2503999/WATCHMEN%20TS%20XVI

255

D%20-%20STG%20PIRATES-COVE.BIZ.html http://alivetorrents.com/torrent/2417409/watchmen-ts-xvid-stg-pirates- cove.biz http://www.kickasstorrents.com/t2448625.html watchmen TS Xvid JamBo http://www.monova.org/details/2511960/WATCHMEN%20TS%20XVI movie shack org D%20-%20JAMBO%20HTTP%3A%2F%2FMOVIE- SHACK.ORG.html http://alivetorrents.com/torrent/2424391/watchmen-ts-xvid-jambo-http:-- movie-shack. http://www.kickasstorrents.com/t2448636.html watchmen TS Xvid JamBo http://www.torrenthound.com/hash/ccbcb8ba803d1391b82d9bca3196733 xtreme scene co uk/ 6e27f2d88/torrent-info/watchmen-TS-Xvid--JamBo http://www.monova.org/details/2514750/WATCHMEN%20TS%20XVI D%20-%20JAMBO%20HTTP%3A%2F%2FXTREME- SCENE.CO.UK%2F.html http://alivetorrents.com/torrent/2426752/watchmen-ts-xvid-jambo-http:-- xtreme-scene.co.uk- http://www.kickasstorrents.com/t2268326.html Watchmen Die Waechter 2009 http://www.monova.org/details/2424745/WATCHMEN%20DIE%20WA TS LD German XviD TRG ECHTER%202009%20TS%20LD%20GERMAN%20XVID- www phantombitz com TRG%20WWW.PHANTOMBITZ.COM.html http://alivetorrents.com/torrent/2353449/watchmen-die-waechter-2009- ts-ld-german-xvid-trg http://www.kickasstorrents.com/t2154458.html Watchmen Die Waechter 2009 http://www.kickasstorrents.com/t2154693.html TS LD German XviD TRG mazysmadhouse net Watchmen Die Waechter 2009 http://www.torrenthound.com/hash/9c27e80baa15cc04cd3ca61dc39de4c TS LD German XviD TRG 2c2a511e4/torrent-info/Watchmen-Die-Waechter-2009-TS-LD-German- mazysnuthouse org XviD-TRG-mazysnuthouse-org http://www.monova.org/details/2424649/WATCHMEN%20DIE%20WA ECHTER%202009%20TS%20LD%20GERMAN%20XVID-TRG- MAZYSNUTHOUSE%20ORG.html http://alivetorrents.com/torrent/2354212/watchmen-die-waechter-2009- ts-ld-german-xvid-trg-mazysnuthouse- http://www.kickasstorrents.com/t2154192.html Watchmen READNFO TS http://www.monova.org/details/2428346/WATCHMEN%20READNFO XviD DEViSE toxictorrents %20TS%20XVID- com DEVISE%20%5BTOXICTORRENTS%20COM%5D.html http://www.kickasstorrents.com/t2451712.html Watchmen READNFO TS http://www.monova.org/details/2428200/WATCHMEN%20READNFO XviD DEViSE www %20TS%20XVID- twilighttorrents net DEVISE%20%20%20HTTP%3A%2F%2FWWW.TWILIGHTTORREN SEEDBOXES!!! TS.NET%20%20%20SEEDBOXES%21%21%21.html http://www.kickasstorrents.com/t2160905.html Watchmen READNFO TS http://www.monova.org/details/2428667/WATCHMEN%20READNFO XviD DEViSE %20TS%20XVID-DEVISE.html http://www.kickasstorrents.com/t2451662.html Watchmen READNFO TS http://www.monova.org/details/2428615/WATCHMEN%20READNFO XviD DEViSE lovethescene %20TS%20XVID-DEVISE%20%5BLOVETHESCENE.NET%5D.html net http://www.kickasstorrents.com/t2159976.html Watchmen READNFO TS http://www.torrenthound.com/hash/ee63c8514677cd913fc87d5eb22c20d XviD DEViSE mazysnuthouse cfb53282d/torrent-info/Watchmen-READNFO-TS-XviD-DEViSE- org mazysnuthouse-org http://www.monova.org/details/2428223/WATCHMEN%20READNFO

256

%20TS%20XVID-DEVISE-MAZYSNUTHOUSE%20ORG.html http://www.kickasstorrents.com/t2159709.html Watchmen READNFO TS http://www.monova.org/details/2428176/WATCHMEN%20READNFO XviD DEViSE %20TS%20XVID-DEVISE%20MAZYSMADHOUSE%20NET.html Mazysmadhouse net http://www.kickasstorrents.com/t2159707.html Watchmen READNFO TS http://www.monova.org/details/2430376/WATCHMEN%20READNFO XviD DEViSE Torrentzilla %20TS%20XVID- org DEVISE%20%5B%20TORRENTZILLA.ORG%20%5D.html http://www.kickasstorrents.com/t2162221.html Watchmen READNFO TS http://www.monova.org/details/2429623/WATCHMEN%20READNFO XviD DEViSE www heavens %20TS%20XVID-DEVISE%5BWWW%20HEAVENS- above biz ABOVE%20BIZ%5D.html http://www.kickasstorrents.com/t2161248.html Watchmen READNFO TS http://www.monova.org/details/2430783/WATCHMEN%20READNFO XviD DEViSE www %20TS%20XVID-DEVISE- darknesstorrents com WWW.DARKNESSTORRENTS.COM.html http://www.kickasstorrents.com/t2164104.html Watchmen READNFO TS http://www.torrenthound.com/hash/f92dd10eddbd9ba07d7d5189dc3fe26 XviD DEViSE www pretome 1b4e5a9df/torrent-info/Watchmen-READNFO-TS-XviD--DEViSE net http://www.monova.org/details/2431053/WATCHMEN%20READNFO %20TS%20XVID- DEVISE%5BWWW%20PRETOME%20NET%5D.html http://www.kickasstorrents.com/t2162976.html Watchmen READNFO TS http://www.monova.org/details/2431119/WATCHMEN%20READNFO XviD DEViSE www %20TS%20XVID-DEVISE%20WWW.PHANTOMBITZ.COM.html phantombitz com http://www.kickasstorrents.com/t2162789.html Watchmen READNFO TS http://www.monova.org/details/2433856/WATCHMEN%20READNFO XviD DEViSE ceasers palace %20TS%20XVID-DEVISE%20CEASERS-PALACE%20INFO.html info http://www.kickasstorrents.com/t2165810.html Watchmen Die Waechter TS http://www.torrenthound.com/hash/065c9a495789c9174a0936cc752df91 LD German READ NFO b232c874d/torrent-info/Watchmen-Die-Waechter-TS-LD-German- XviD ADLER READ-NFO-XviD-ADLER http://www.monova.org/details/2433945/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN%20READ%20NFO%20XVID- ADLER.html http://torrentportal.com/details/4027524/Watchmen.Die.Waechter.TS.LD .German.READ.NFO.XviD-ADLER.torrent http://www.kickasstorrents.com/t2166177.html Watchmen 2009 TS XVID http://www.torrenthound.com/hash/cb026a7889fbff386ea81a9191eaaf8e www demonpeer org bd545f17/torrent-info/Watchmen-2009-TS-XVID http://www.monova.org/details/2437503/WATCHMEN%202009%20TS %20XVID%20WWW%20DEMONPEER%20ORG.html http://alivetorrents.com/torrent/2364770/watchmen-2009-ts-xvid-- http://www.kickasstorrents.com/t2451589.html MoR Watchmen Die Waechter http://www.torrenthound.com/hash/6fd9a2a4df9e434edcb7e07aeb956866 TS LD German XviD TRG 2 c7ff64d7/torrent-info/MoR-Watchmen-Die-Waechter-TS-LD-German- XviD-TRG-2 http://www.monova.org/details/2448123/%5BMOR%5DWATCHMEN% 20DIE%20WAECHTER%20TS%20LD%20GERMAN%20XVID- TRG%282%29.html http://alivetorrents.com/torrent/2372884/morwatchmen-die-waechter-ts- ld-german-xvid-trg2 http://www.kickasstorrents.com/t2181734.html Watchmen Die Waechter TS http://www.torrenthound.com/hash/1c8763bfe2825a64bf49c8d6de59c2a2 LD German 4b2dde75/torrent-info/Watchmen-Die-Waechter-TS-LD-German

257

http://www.monova.org/details/2523657/WATCHMEN%20DIE%20WA ECHTER%20TS%20LD%20GERMAN.html http://alivetorrents.com/torrent/2434247/watchmen-die-waechter-ts-ld- german http://www.kickasstorrents.com/t2280075.html Watchmen CAM XViD http://1337x.org/torrent/14063/0/ CAMERA www Yestorrent http://www.vertor.com/torrents/687271/Watchmen-CAM-XViD- com 1921 14277 CAMERA-www-Yestorrent-com-1921%2F14277 http://www.h33t.com/details.php?id=4f8fd97310857d694085ac4ff6da2e7 930027fed http://btjunkie.org/torrent/Watchmen-CAM-XViD-CAMERA- ESeeds/44864f8fd97310857d694085ac4ff6da2e7930027fed http://www.torrenthound.com/hash/4f8fd97310857d694085ac4ff6da2e79 30027fed/torrent-info/Watchmen-CAM-XViD-CAMERA-ENGLISH- AUDIO-NoRar%C3%82%C2%A4 http://www.yourbittorrent.com/torrent/113048/Watchmen.CAM.XViD- CAMERA.html http://www.monova.org/details/2425354/WATCHMEN%20CAM%20X VID-CAMERA%20ENGLISH%20AUDIO-NORAR.html http://www.fulldls.com/torrent-movies-1598516.html http://alivetorrents.com/torrent/2354736/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2156465.html Watchmen CAM XViD http://thepiratebay.org/torrent/4758270/Watchmen.CAM.XViD-700MB 700MB http://www.torrenthound.com/hash/35684b1c4bcf20773c0b255af6974c6 8899b4aeb/torrent-info/Watchmen-CAM-XViD-700MB http://www.monova.org/details/2430104/WATCHMEN%202009%20VE RY%20GOOD%20CAM.html http://alivetorrents.com/torrent/2358608/watchmen-2009-very-good-cam http://www.kickasstorrents.com/t2161654.html Watchmen 2009 Eng CAM http://thepiratebay.org/torrent/4757567/Watchmen.[2009.Eng].CAM.Div DivX LTT X-LTT http://btjunkie.org/torrent/Watchmen-2009-Eng-CAM-DivX- LTT/44957f7bd24b9fa21c77a42b2403e93171ce69473ecf http://www.torrenthound.com/hash/7f7bd24b9fa21c77a42b2403e93171c e69473ecf/torrent-info/Watchmen-2009-Eng-CAM-DivX-LTT http://www.yourbittorrent.com/torrent/113023/Watchmen.%5B2009.Eng %5D.CAM.DivX-LTT.html http://www.monova.org/details/2428198/WATCHMEN%20%5B2009%2 0ENG%5D%20CAM%20DIVX-LTT.html http://www.fulldls.com/torrent-movies-1670121.html http://alivetorrents.com/torrent/2480356/isohunt-watchmen-2009-eng- cam-divx-ltt-4757567-tpb http://www.kickasstorrents.com/t2159713.html Watchmen CAM XViD http://www.torrenthound.com/hash/fb55715475bda0f7c86209ba91d3019 CAMERA scenepirates com afd7c2ed0/torrent-info/Watchmen-CAM-XViD-CAMERA-ENGLISH- AUDIO http://www.yourbittorrent.com/torrent/12098/Watchmen.CAM.XViD- CAMERA.html http://www.monova.org/details/2425429/WATCHMEN%20CAM%20X VID-CAMERA%28SCENEPIRATES.COM%29.html http://www.fulldls.com/torrent-movies-1598531.html http://alivetorrents.com/torrent/2354661/watchmen-cam-xvid- camerascenepirates. http://www.kickasstorrents.com/t2156505.html Watchmen 2009 TS XViD http://thepiratebay.org/torrent/4758433/Watchmen_(2009)_TS.XViD- CAMERA CAMERA

258

http://btjunkie.org/torrent/Demonoid-Watchmen-2009-TS-XViD- CAMERA/2834db1b126df4c1712dcab03c6f9ef9740dbb302e20 http://www.torrenthound.com/hash/db1b126df4c1712dcab03c6f9ef9740d bb302e20/torrent-info/Watchmen-2009-TS-XViD-CAMERA http://www.btmon.com/Video/Unsorted/Demonoid_Watchmen_2009_TS .XViD-CAMERA.torrent.html http://www.monova.org/details/2430551/%5BDEMONOID%5D%20WA TCHMEN%20%282009%29%20TS.XVID-CAMERA.html http://www.fulldls.com/torrent-movies-1600883.html http://alivetorrents.com/torrent/2359133/demonoid-watchmen-2009- ts.xvid-camera http://www.kickasstorrents.com/t2162193.html WATCHMEN cam recode http://www.vertor.com/torrents/693800/WATCHMEN_cam_recode_sido sidonkey nwrg avi nkey%28nwrg%29-avi http://www.torrenthound.com/hash/b48e505d971db335282df9bacc33768 fe33f512d/torrent-info/WATCHMEN-cam-recode-sidonkey-nwrg-avi http://www.btmon.com/Video/Movies/WATCHMEN_cam_recode_sidon key_nwrg_.avi.torrent.html http://www.monova.org/details/2443141/WATCHMEN_CAM_RECOD E_SIDONKEY%28NWRG%29.AVI%09.html http://torrentportal.com/details/4034205/WATCHMEN_cam_recode_sid onkey%28nwrg%29.avi.torrent http://alivetorrents.com/torrent/5602585/watchmen_cam_recode_sidonke ynwrg.avi http://www.kickasstorrents.com/t2176518.html Watchmen CAM XViD http://btjunkie.org/torrent/Watchmen-CAM-XViD-CAMERA-Www- CAMERA Www Overget Overget-Com/45198d23e648b49db81db973c7f56e5114a1acb7d5de Com http://www.torrenthound.com/hash/8d23e648b49db81db973c7f56e5114a 1acb7d5de/torrent-info/Watchmen-CAM-XViD-CAMERA http://www.monova.org/details/2428677/WATCHMEN%20CAM%20X VID-CAMERA%5BWWW%20OVERGET%20COM%5D.html http://torrentportal.com/details/4024316/Watchmen+CAM+XViD- CAMERA%5BWww.Overget.Com%5D.torrent http://www.fulldls.com/torrent-movies-1600127.html http://alivetorrents.com/torrent/2360019/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2160362.html http://www.torrentbit.nl/torrent/248841/Watchmen+CAM+XViD- CAMERA/ Watchmen CAM DivX http://www.monova.org/details/2432480/WATCHMEN%20CAM%20DI BLaCKMiKE avi VX-BLACKMIKE%20AVI.html http://alivetorrents.com/torrent/2360978/watchmen-cam-divx-blackmike- avi http://www.kickasstorrents.com/t2164719.html Watchmen CAM XViD http://www.monova.org/details/2425255/WATCHMEN%20CAM%20X CAMERA VID-CAMERA.html http://alivetorrents.com/torrent/2354535/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2156292.html Watchmen CAM XViD http://www.monova.org/details/2427962/WATCHMEN%20CAM%20X CAMERA FilmsBT com VID-CAMERA%20%5BFILMSBT%20COM%5D.html http://www.fulldls.com/torrent-movies-1599800.html http://alivetorrents.com/torrent/2357188/watchmen-cam-xvid-camera- filmsbt- http://www.kickasstorrents.com/t2159416.html Watchmen CAM 2009 http://www.vertor.com/torrents/690610/Watchmen-2009-cam- desifilmz com goodquality-desifilmz-385%2F784 http://thepiratebay.org/torrent/4756591/Watchmen.2009.cam.goodquality

259

.desifilmz http://btjunkie.org/torrent/Watchmen-2009-cam-goodquality- desifilmz/395223883426bcf276f529ef6eca7b753bb2a9f05f6e http://www.torrenthound.com/hash/23883426bcf276f529ef6eca7b753bb2 a9f05f6e/torrent-info/Watchmen-2009-cam-goodquality-desifilmz http://www.monova.org/details/2426092/WATCHMEN%20CAM%2020 09%20DESIFILMZ%20COM.html http://alivetorrents.com/torrent/2355487/watchmen-cam-2009-desifilmz- http://www.kickasstorrents.com/t2157298.html Watchmen FRENCH 2009 http://www.torrenthound.com/hash/8f02ec597d642209749fe005b7a8567 screener cam TS MD e7555242e/torrent-info/Watchmen-FRENCH-2009-screener-cam-TS- REPACK 1CD XViD STS MD-REPACK-1CD-XViD-STS-avi www miragetorrent com avi http://www.monova.org/details/2424914/WATCHMEN%20FRENCH%2 0TS%20MD%20REPACK%201CD%20XVID- STS%20%28%20%20WWW.MIRAGETORRENT.COM%20%20%29% 20AVI.html http://alivetorrents.com/torrent/2519559/watchmen-french-2009- screener-cam-ts-md-repack-1cd-xvid-sts.miragetorrent.-avi http://www.kickasstorrents.com/t3005616.html Watchmen TS Xvid Camera http://www.monova.org/details/2429839/WATCHMEN-TS-XVID- 2cds ESPATORRENTS com CAMERA-2CDS-%5BESPATORRENTS%20COM%5D.html http://alivetorrents.com/torrent/2358386/watchmen-ts-xvid-camera-2cds- espatorrents- http://www.kickasstorrents.com/t2161378.html Watchmen TS XViD CAMERA Watchmen 2009 Cam Xdiv http://www.monova.org/details/2437782/WATCHMEN- Subtitulado www torrent bt 2009%20CAM%20XDIV%20SUBTITULADO%20WWW%20TORRE com ar NT-BT%20COM%20AR.html http://www.kickasstorrents.com/t2170452.html Watchmen 2009 iTALiAN http://www.torrenthound.com/hash/280db9096a295068c3e22e93366e616 iNTERNAL READNFO LD 3acc46c43/torrent-info/Watchmen-2009-iTALiAN-iNTERNAL- CAM XviD SiLENT READNFO-LD-CAM-XviD-SiLENT http://www.monova.org/details/2487796/WATCHMEN.2009.ITALIAN.I NTERNAL.READNFO.LD.CAM.XVID-SILENT.html http://torrentportal.com/details/4062340/Watchmen.2009.iTALiAN.iNTE RNAL.READNFO.LD.CAM.XviD-SiLENT.torrent http://www.fulldls.com/torrent-others-1627306.html http://www.kickasstorrents.com/t2229075.html Watchmen CAM XViD http://www.torrenthound.com/hash/3684d5378ed0033822350060a02da5 CAMERA PeerDen com 9ac25dc66a/torrent-info/Watchmen-CAM-XViD--CAMERA--PeerDen- com http://www.btmon.com/Video/Unsorted/Watchmen_CAM_XViD- CAMERA_-_PeerDen_com.torrent.html http://www.monova.org/details/2425427/WATCHMEN%20CAM%20X VID-CAMERA%20-%20PEERDEN%20COM.html http://www.fulldls.com/torrent-movies-1598527.html http://alivetorrents.com/torrent/2354674/watchmen-cam-xvid-camera- peerden- http://www.kickasstorrents.com/t2156850.html Watchmen CAM XViD http://www.newtorrents.info/torrent/3933/Watchmen.CAM.XViD- CAMERA CAMERA-.html?nopop=1 http://www.btmon.com/Video/Unsorted/Watchmen_CAM_XViD- CAMERA-_tracker_BTARENA_org.torrent.html http://www.monova.org/details/2425352/WATCHMEN%20CAM%20X

260

VID-CAMERA-%5BTRACKER%20BTARENA%20ORG%5D.html http://www.fulldls.com/torrent-movies-1598534.html http://alivetorrents.com/torrent/2354658/watchmen-cam-xvid-camera- tracker-btarena- http://www.kickasstorrents.com/t2156437.html Watchmen CAM XViD http://www.torrenthound.com/hash/a4c9ba67549bcfb5dcc618ea3e3c0995 CAMERA avi cb16c5c6/torrent-info/Watchmen-CAM-XViD--CAMERA-avi http://www.monova.org/details/2425652/WATCHMEN%20CAM%20X VID-CAMERA%20AVI.html http://alivetorrents.com/torrent/2354965/watchmen-cam-xvid-camera-avi http://www.kickasstorrents.com/t2156780.html Watchmen CAM XViD http://www.torrenthound.com/hash/107ff1fc52864dc65ac46195029bceec CAMERA www torrent loco 18672a24/torrent-info/Watchmen-CAM-XViD-CAMERA-ar com ar http://www.monova.org/details/2425593/WATCHMEN%20CAM%20X VID-CAMERA%20WWW.TORRENT-LOCO.COM.AR.html http://torrentportal.com/details/4022005/Watchmen.CAM.XViD- CAMERA.torrent-loco.torrent http://www.fulldls.com/torrent-others-1599225.html http://alivetorrents.com/torrent/2354884/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2156640.html Watchmen CAM CD2 XViD http://www.monova.org/details/2427089/WATCHMEN%20CAM%20C CAMERA D2%20XVID-CAMERA.html http://alivetorrents.com/torrent/2356476/watchmen-cam-cd2-xvid-camera http://www.kickasstorrents.com/t2451734.html Watchmen CAM CD1 XViD http://www.monova.org/details/2427087/WATCHMEN%20CAM%20C CAMERA D1%20XVID-CAMERA.html http://alivetorrents.com/torrent/2356467/watchmen-cam-cd1-xvid-camera http://www.kickasstorrents.com/t2451745.html Watchmen CAM XViD http://www.monova.org/details/2428469/WATCHMEN%20CAM%20X CAMERA toxictorrents com VID-CAMERA%20%5BTOXICTORRENTS%20COM%5D.html http://alivetorrents.com/torrent/2357502/watchmen-cam-xvid-camera- toxictorrents- http://www.kickasstorrents.com/t2451723.html watchmen cam sample re http://www.monova.org/details/2427648/WATCHMEN%20CAM%20S encoded avi AMPLE%20RE-ENCODED.AVI.html http://www.fulldls.com/torrent-movies-1599626.html http://alivetorrents.com/torrent/2356950/watchmen-cam-sample-reencoded.avi http://www.kickasstorrents.com/t2158838.html Watchmen CAM XViD http://www.torrenthound.com/hash/1a00d4981356e448756d14a5bf2c1a6 c604a9de9/torrent-info/Watchmen-CAM-XViD http://www.monova.org/details/2454187/WATCHMEN%20CAM%20X VID.html http://alivetorrents.com/torrent/2377358/watchmen-cam-xvid http://www.kickasstorrents.com/t2188035.html Watchmen 2009 CAM http://www.torrenthound.com/hash/685fe0664178c5e5b455dd600320c24 CAMERA Kingdom KvCD by 2950fcc95/torrent-info/Watchmen-2009-CAM--CAMERA-Kingdom-- Bexley KvCD-by-Bexley http://www.monova.org/details/2425541/WATCHMEN%202009%20CA M-CAMERA%20%28KINGDOM- KVCD%20BY%20BEXLEY%29.html http://alivetorrents.com/torrent/2354776/watchmen-2009-cam-camera- kingdom-kvcd-by-bexley http://www.kickasstorrents.com/t2156608.html Watchmen TS XViD http://www.vertor.com/torrents/1213599/Watchmen-TS-XViD-

261

CAMERA www torrent loco CAMERA-torrent-loco-com-ar-154%2F632 com ar http://www.torrenthound.com/hash/90117b317f537c765186e7299e13164 3cbf004d5/torrent-info/Watchmen-TS-XViD-CAMERA-ar http://www.yourbittorrent.com/torrent/766558/Watchmen.TS.XViD- CAMERA.www.torrent-loco.com.ar.html http://www.monova.org/details/2429997/WATCHMEN%20TS%20XVI D-CAMERA%20WWW%20TORRENT-LOCO%20COM%20AR.html http://torrentportal.com/details/4024809/Watchmen.TS.XViD- CAMERA.www.torrent-loco.com.ar.torrent http://alivetorrents.com/torrent/2358586/watchmen-ts-xvid-camera---ar http://www.kickasstorrents.com/t2161594.html Watchmen TS XViD http://thepiratebay.org/torrent/4758727/Watchmen.TS.XViD-CAMERA CAMERA http://www.torrenthound.com/hash/3966f01dc21bb234618e8fa522801f4 3a2d9a011/torrent-info/Watchmen-TS-XViD-CAMERA-FilmsBT-com http://www.monova.org/details/2431422/WATCHMEN.TS.XVID- CAMERA.%5BFILMSBT.COM%5D%09.html http://torrentportal.com/details/4027867/Watchmen.TS.XViD- CAMERA.%5BFilmsBT.com%5D.torrent http://alivetorrents.com/torrent/5608860/watchmen.ts.xvid-camera http://www.kickasstorrents.com/t2163462.html Watchmen Tntvillage http://www.vertor.com/torrents/1439671/Watchmen-Scambioetico Scambioetico http://www.torrenthound.com/hash/261697716b0f17ecbbe4672e1a165f6 3f5bd6dbd/torrent-info/Watchmen-Tntvillage-Scambioetico http://www.btmon.com/Other/Unsorted/ReleaseFumetti_Watchmen.cbz.t orrent.html http://www.monova.org/details/3081950/WATCHMEN%20%5BTNTVI LLAGE.SCAMBIOETICO%5D.html http://torrentportal.com/details/4331641/Watchmen.cbz.torrent http://www.fulldls.com/torrent-ebooks-1904476.html http://extratorrent.com/torrent/2132228/Watchmen+Scambioetico.html http://alivetorrents.com/torrent/2917245/watchmen- tntvillage.scambioetico http://www.kickasstorrents.com/t2957785.html Watchmen CAM XViD http://www.monova.org/details/2425260/WATCHMEN%20CAM%20X CAMERA ENGLISH AUDIO VID- www supreme speeds com CAMERA%20%20ENGLISH%20AUDIO%20WWW%20SUPREME- SPEEDS%20COM.html http://alivetorrents.com/torrent/2354567/watchmen-cam-xvid-camera-- english-audio-- http://www.kickasstorrents.com/t2156313.html Watchmen CAM XViD http://www.monova.org/details/2425246/WATCHMEN%20CAM%20X CAMERA www heavens VID-CAMERA%5BWWW%20HEAVENS-ABOVE%20BIZ%5D.html above biz http://alivetorrents.com/torrent/2354555/watchmen-cam-xvid-camera-biz http://www.kickasstorrents.com/t2156302.html Watchmen CAM XViD http://www.monova.org/details/2425432/WATCHMEN%20CAM%20X CAMERA www seedmore org VID-CAMERA%5BWWW%20SEEDMORE%20ORG%5D.html http://www.fulldls.com/torrent-movies-1598517.html http://alivetorrents.com/torrent/2354730/watchmen-cam-xvid-camera- http://www.kickasstorrents.com/t2156528.html Watchmen CAM XViD http://www.yourbittorrent.com/torrent/302708/Watchmen.CAM.XViD- CAMERA CAMERA.html http://www.monova.org/details/2425340/WATCHMEN%20CAM%20X VID-CAMERA-%5BTRACKER%20BTARENA%20ORG%5D.html http://www.fulldls.com/torrent-movies-1598535.html http://alivetorrents.com/torrent/2354650/watchmen-cam-xvid-camera- tracker-btarena-

262

http://www.kickasstorrents.com/t2156451.html Watchmen CAM XViD http://www.monova.org/details/2425426/WATCHMEN%20CAM%20X CAMERA megaups com VID-CAMERA%20%5BMEGAUPS.COM%5D.html http://alivetorrents.com/torrent/2354603/watchmen-cam-xvid-camera- megaups. http://www.kickasstorrents.com/t2156408.html Watchmen CAM XViD http://www.monova.org/details/2425341/WATCHMEN%20CAM%20X CAMERA www VID- twilighttorrents net/ CAMERA%20%20%20HTTP%3A%2F%2FWWW.TWILIGHTTORRE SEEDBOXES NTS.NET%2F%20%20%20%20SEEDBOXES.html http://alivetorrents.com/torrent/2354602/watchmen-cam-xvid-camera--- seedboxes http://www.kickasstorrents.com/t2156407.html Watchmen CAM XViD http://www.monova.org/details/2425602/WATCHMEN%20CAM%20X CAMERA mazysnuthouse org VID-CAMERA%20MAZYSNUTHOUSE%20ORG.html http://alivetorrents.com/torrent/2354785/watchmen-cam-xvid-camera- mazysnuthouse- http://www.kickasstorrents.com/t2156590.html Watchmen CAM XViD http://www.monova.org/details/2425519/WATCHMEN%20CAM%20X CAMERA www pretome net VID-CAMERA%5BWWW%20PRETOME%20NET%5D.html http://extratorrent.com/torrent/1780773/Watchmen+CAM+XViD- CAMERA+net.html http://alivetorrents.com/torrent/2354784/watchmen-cam-xvid-camera- http://www.kickasstorrents.com/t2156614.html www torrent god com http://www.monova.org/details/2425888/%5BWWW%20TORRENT- Watchmen CAM XViD GOD%20COM%5D%20-%20WATCHMEN%20CAM%20XVID- CAMERA CAMERA.html http://alivetorrents.com/torrent/2355080/--watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2156939.html Watchmen CAM XViD http://www.monova.org/details/2426192/WATCHMEN.CAM.XVID- CAMERA www sparksden CAMERA%20%20-%20%20WWW.SPARKSDEN.ORG.html org http://extratorrent.com/torrent/1781193/Watchmen.CAM.XViD- CAMERA++-.html http://alivetorrents.com/torrent/2355452/watchmen.cam.xvid-camera- http://www.kickasstorrents.com/t2157199.html Watchmen CAM XViD http://www.monova.org/details/2426673/WATCHMEN%20CAM%20X CAMERA VID-CAMERA.html http://www.fulldls.com/torrent-others-1599376.html http://alivetorrents.com/torrent/2356111/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2158693.html Watchmen CAM XViD http://www.monova.org/details/2428012/WATCHMEN%20CAM%20X CAMERA movie shack org VID-CAMERA%20HTTP%3A%2F%2FMOVIE-SHACK.ORG.html http://alivetorrents.com/torrent/2357126/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2159271.html Watchmen CAM XViD http://www.monova.org/details/2428446/WATCHMEN%20CAM%20X CAMERA Sample And NFO VID-CAMERA- toxictorrents com SAMPLE%20AND%20NFO%20%5BTOXICTORRENTS%20COM%5 D.html http://alivetorrents.com/torrent/2357503/watchmen-cam-xvid-camera- sample-and-nfo-toxictorrents- http://www.kickasstorrents.com/t2451719.html Watchmen CAM XViD http://www.monova.org/details/2429206/WATCHMEN.CAM.XVID- CAMERA torrent www twisty CAMERA.TORRENT%28WWW.TWISTY-MISTYS.COM%29.html mistys com http://www.fulldls.com/torrent-movies-1600266.html http://alivetorrents.com/torrent/2358132/watchmen.cam.xvid-

263

camera.torrent http://www.kickasstorrents.com/t2160812.html Watchmen CAM XViD http://www.monova.org/details/2430311/WATCHMEN%20CAM%20X CAMERA Torrentzilla org VID-CAMERA%20%5B%20TORRENTZILLA.ORG%20%5D.html http://www.kickasstorrents.com/t2164522.html Watchmen CAM XViD http://www.monova.org/details/2430930/WATCHMEN%20CAM%20X CAMERA www phantombitz VID-CAMERA%20WWW.PHANTOMBITZ.COM.html com http://alivetorrents.com/torrent/2359565/watchmen-cam-xvid-camera http://www.kickasstorrents.com/t2162881.html Watchmen CAM XViD http://www.torrenthound.com/hash/8cd365f51b0a138445ab78e2f1c5d3b CAMERA www 68ac57420/torrent-info/Watchmen-CAM-XViD--CAMERA darknesstorrents com http://www.monova.org/details/2430811/WATCHMEN%20CAM%20X VID-CAMERA-WWW.DARKNESSTORRENTS.COM.html http://alivetorrents.com/torrent/2359391/watchmen-cam-xvid-camera- http://www.kickasstorrents.com/t2164106.html Watchmen CAM XViD http://www.monova.org/details/2431641/WATCHMEN%20CAM%20X CAMERA www torrentsforall VID-CAMERA%20%5BWWW.TORRENTSFORALL.NET%5D.html net http://www.fulldls.com/torrent-movies-1601395.html http://alivetorrents.com/torrent/2360172/watchmen-cam-xvid-camera- http://www.kickasstorrents.com/t2163741.html Watchmen CAM XViD http://www.monova.org/details/2434169/WATCHMEN%20CAM%20X CAMERA ceasers palace info VID-CAMERA%20CEASERS-PALACE%20INFO.html http://alivetorrents.com/torrent/2361640/watchmen-cam-xvid-camera- ceasers-palace- http://www.kickasstorrents.com/t2165808.html Watchmen CAM XViD ENG http://www.torrenthound.com/hash/ae28a8d9fc68cb51cd03be5ee413650e www demonpeer org 5c947385/torrent-info/Watchmen-CAM-XViD http://www.monova.org/details/2425549/WATCHMEN%20CAM%20X VID%20WWW%20DEMONPEER%20ORG.html http://alivetorrents.com/torrent/2354889/watchmen-cam-xvid-eng-- http://www.kickasstorrents.com/t2156694.html watchmen cam rar http://www.monova.org/details/2427058/WATCHMEN%20CAM%20R AR.html http://alivetorrents.com/torrent/2356321/watchmen-cam-rar http://www.kickasstorrents.com/t2159721.html Watchmen TS XViD http://www.vertor.com/torrents/1984261/Watchmen-TS-XViD- CAMERA scenepirates com CAMERA%28scenepirates-com%29 http://www.monova.org/details/2430440/WATCHMEN%20TS%20XVI D-CAMERA%28SCENEPIRATES.COM%29.html http://alivetorrents.com/torrent/2359094/watchmen-ts-xvid- camerascenepirates. http://www.kickasstorrents.com/t2162054.html Watchmen TS XViD http://www.torrenthound.com/hash/e731530d4ae5f0ca839d4ac65328fd8d CAMERA mazysnuthouse org 3f566f06/torrent-info/Watchmen-TS-XViD-CAMERA-mazysnuthouse- org http://www.monova.org/details/2429902/WATCHMEN%20TS%20XVI D-CAMERA%20MAZYSNUTHOUSE%20ORG.html http://alivetorrents.com/torrent/2358414/watchmen-ts-xvid-camera- mazysnuthouse- http://www.kickasstorrents.com/t2161430.html Watchmen TS XViD http://www.torrenthound.com/hash/b53ff79047f4bff7f8953612b911e73e CAMERA www supreme 95da2089/torrent-info/Watchmen-TS-XViD--CAMERA speeds com http://www.monova.org/details/2429909/WATCHMEN%20TS%20XVI D-CAMERA%20WWW%20SUPREME-SPEEDS%20COM.html http://alivetorrents.com/torrent/2358404/watchmen-ts-xvid-camera--

264

http://www.kickasstorrents.com/t2161807.html Watchmen TS XViD http://www.monova.org/details/2430979/WATCHMEN%20TS%20XVI CAMERA,mazysmadhouse D-CAMERA%2CMAZYSMADHOUSE%20NET.html net http://www.kickasstorrents.com/t2162894.html Watchmen TS XViD http://www.torrenthound.com/hash/615ed5a13368fb139531df7a9e27548 CAMERA www pretome net 112bd4d2b/torrent-info/Watchmen-TS-XViD--CAMERA http://www.monova.org/details/2431130/WATCHMEN%20TS%20XVI D-CAMERA%5BWWW%20PRETOME%20NET%5D.html http://www.fulldls.com/torrent-movies-1601123.html http://alivetorrents.com/torrent/2359670/watchmen-ts-xvid-camera- http://www.kickasstorrents.com/t2162992.html Watchmen TS XViD http://www.monova.org/details/2431824/WATCHMEN%20TS%20XVI CAMERA www heavens D-CAMERA%5BWWW%20HEAVENS-ABOVE%20BIZ%5D.html above biz http://alivetorrents.com/torrent/2360228/watchmen-ts-xvid-camera-biz http://www.kickasstorrents.com/t2163856.html Watchmen TS XViD http://www.monova.org/details/2431761/WATCHMEN%20TS%20XVI CAMERA Torrentzilla org D-CAMERA%20%5B%20TORRENTZILLA.ORG%20%5D.html http://www.kickasstorrents.com/t2163941.html Watchmen TS XViD http://www.monova.org/details/2431488/WATCHMEN%20TS%20XVI CAMERA teamfx org D-CAMERA-%5BTEAMFX%20ORG%5D.html http://alivetorrents.com/torrent/2359991/watchmen-ts-xvid-camera- teamfx- http://www.kickasstorrents.com/t2161617.html Watchmen TS/CAM http://thepiratebay.org/torrent/4759666/Watchmen_TS_CAM http://www.torrenthound.com/hash/207c971eeef26b0a1ee3f1bf52786011 0ea7e53e/torrent-info/Watchmen-TS-CAM http://www.monova.org/details/2434487/WATCHMEN%20TS%2FCAM %09.html http://alivetorrents.com/torrent/5608063/watchmen-ts-cam http://www.kickasstorrents.com/t2165734.html Watchmen TS XViD http://www.torrenthound.com/hash/dbd11880a87eb47b4e1aab73887b5dd CAMERA www torrentsforall f4af547da/torrent-info/Watchmen-TS-XViD-CAMERA net http://www.monova.org/details/2454541/WATCHMEN%20TS%20XVI D-CAMERA%20%5BWWW.TORRENTSFORALL.NET%5D.html http://alivetorrents.com/torrent/2377639/watchmen-ts-xvid-camera- .torrentsforall. http://www.kickasstorrents.com/t2188374.html

265

REFERENCES

[1] Fei-Yue Wang, Guanpi Lai and Shuming Tang, "An application specific knowledge engine for researches in Intelligent Transportation Systems," in IEEE Conference on Intelligent Transportation Systems Washington, D.C., 2004, pp. 841 - 846. [2] Jialun Qin, Yilu Zhou, Edna Reid, Guanpi Lai and Hsinchun Chen, "Analyzing terror campaigns on the internet: Technical sophistication, content richness, and Web interactivity," International journal of human-computer studies, vol. 65, pp. 71 - 84, 2007. [3] Phil Bernstein, Michael Brodie, Stefano Ceri, David DeWitt, Mike Franklin, Hector Garcia-Molina, Jim Gray, Jerry Held, Joe Hellerstein, H. V. Jagadish, Michael Lesk, Dave Maier, Je Naughton, Hamid Pirahesh, Mike Stonebraker and Je Ullman, "The asilomar report on database research," ACM Sigmod Record, vol. 27, 1998. [4] Alvin Toffle, Future Shock : Random House, 1970. [5] Thi Truong Avrahami, Lawrence Yau, Luo Si and Jamie Callan, "The FedLemur project: Federated search in the real world," Journal of the American Society for Information Science and Technology, vol. 57, 2006. [6] William McKnight, "Building Business Intelligence: Text Data Mining in Business Intelligence," in Information Management Magazine , 2005 pp. 21 - 22. [7] Robert L. Read and Donald S. Fussell, "A multi-resolution relational data model," in Proc. 18th Int. Conf. Very Large Data Bases , Vancouver, Canada, 1992, pp. 139-150. [8] Justin Langseth, "Clarabridge and Business Objects Link Worlds of Text Mining and Business Intelligence; Partnership Allows Business Users to Access and Analyze Unstructured Data for Improved Enterprise Information Management," in Business Wire , 2006. [9] Robert M Losee, "Browsing mixed structured and unstructured data," Information Processing and Management, vol. 42, pp. 440 - 452, 2006. [10] Martin De Saulles, "Information literacy amongst UK SMEs: an information policy gap," Aslib Proceedings, vol. 59, pp. 68 - 79, 2007. [11] "Powerset - the natural language processing search engine empowered by Hbase in Hadoop, " http://web2innovations.com/money/2007/12/04/powerset- %E2%80%93-the-natural-language-processing-search-engine-empowered-by- hbase-in-hadoop/ , 2007. [12] J. Budzik and K. Hammond, "User interactions with everyday applications as context for just-in-time information access," in Proceedings of the 2000 International Conference on Intelligent User Interfaces (IUI2000) , 2000, pp. 44 - 51. [13] "Atomz Site Search, " http://www.atomz.com/ , 2010. [14] UM Fayyad, G. Piatetsky-Shapiro and P. Smyth, Advances in knowledge discovery and datamining . Cambridge: AAAI Press/MIT Press, 1996. [15] J. R. Quinlan, "Induction of decision trees," Machine Learning, vol. 1, pp. 81 - 106, 1986.

266

[16] J. R. Quinlan, Programs for machine learning . California: Morgan Kaufmann, 1993. [17] P. Clark and T. Niblett, "The CN2 Induction Algorithm," Machine Learning, vol. 3, pp. 261 - 283, 1989. [18] B. Boutsinas, G. Antzoulatos and P. Alevizos, "A novel classification algorithm based on clustering," in First international conference "From scientific computing to computational engineering" , Athens, Greece, 2004. [19] L. Breinman, J. H. Friedman, R. A. Olshen and C. J. Stone, Classification and regression trees . California: Wadsworth and Brooks, 1984. [20] A. A. Freitas, "A survey of evolutionary algorithms for data mining and knowledge discovery," in Advances in evolutionary computation , A. Ghosh and S. Tsutsui, Eds. Berlin: Springer, 2002. [21] J. H. Friedman, "Multivariate adaptive regression splines," Annals of Statistics, vol. 19, 1991. [22] W. Buntine, "Graphical models for discovering knowledge," in Advances in knowledge discovery and data mining , U. M. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, Eds. Menlo Park, CA, USA: American Association for Artificial Intelligence, 1996, pp. 59 - 82. [23] D. E. Rumelhart, G. E. Hinton and R.J. Williams, "Learning internal representations by error propagation," in Parallel distributed processing: explorations in the microstructure of cognition , D. E. Rumelhart and J. L. McClelland, Eds. Cambridge: MIT Press, 1986, pp. 318 - 362. [24] S. Cost and S. Salzberg, "A Weighted nearest neighbour algorithm for learning with symbolic features," Machine Learning, vol. 10, pp. 57 - 78, 1993. [25] S. Dzeroski, "Inductive logic programming and knowledge discovery in databases," in Advances in knowledge discovery and data mining , U. M. Fayyad, G. Piatetsky-Shapiro, and P. Smyth, Eds. Menlo Park, CA, USA: American Association for Artificial Intelligence, 1996, pp. 117 - 152. [26] S. Muggleton and L. Raedt, "Inductive logic programming: theory and methods," Journal of Logic Programming, vol. 19, pp. 629 - 679, 1994. [27] B. Boutsinas and M.N. Vrahatis, "Artificial nonmonotonic neural networks," Artificial Intelligence, vol. 132, pp. 1 - 38, 2001. [28] V.N. Vapnik, Statistical learning theory . New York: Wiley, 1998. [29] V.N. Vapnik, The nature of statistical learning theory . New York: Springer, 2000. [30] V. Vovk, A. Gammerman and G. Shafer, Algorithmic learning in a random world . New York: Springer, 2005. [31] N. Friedman, D. Geiger, M. Goldsmidt, G. Provan, P. Langley and P. Smyth, "Bayesian network classifier," Machine Learning, vol. 29, pp. 131 - 163, 1997. [32] "Digg, " http://digg.com , 2010. [33] "Flickr, " http://www.flickr.com , 2010. [34] "Twitter, " http://www.twitter.com , 2010. [35] Michael K. Bergman, "White Paper: The Deep Web: Surfacing Hidden Value," The Journal of Electronic Publishing vol. 7, August 2001. [36] "Invisible Web gets deeper, " http://searchenginewatch.com/2162871 , August 2000.

267

[37] John Markoff, "How Many Pages in Google? Take a Guess " in The New York Times , September 2005. [38] Lee Ratzan, "Mining the Deep Web: Search strategies that work," in Computer World , December 2006. [39] "ODP – Open Directory Project, " http://www.dmoz.org/ , 2010. [40] "Computer and Information Science Papers CiteSeer Publications ResearchIndex, " http://citeseer.ist.psu.edu/ , 2010. [41] L. von Ahn and L. Dabbish, "Labeling Images with a Computer Game " in Proc. of ACM CHI , 2004 [42] Alex Wright, "Searching the Deep Web - While the Semantic Web may be a long time coming, Deep Web search strategies offer the promise of a semantic Web.," Communications of the ACM, vol. 51, pp. 14 - 15, 2008. [43] "Web 2.0 Suicide Machine, " http://www.suicidemachine.org/ , 2010. [44] "Online communities: Networks that nurture long-distance relationships and local ties. Pew Internet and American Life Project, " http://www.pewinternet.org/reports/toc.asp?Report=47 , 2001. [45] S. E. Bressler and C.E. Grantham, Communities of commerce: Building Internet business communities to accelerate growth, minimize risk, and increase customer loyalty . New York: McGraw-Hill, 2000. [46] "Participation Inequality: Encouraging More Users to Contribute, " http://www.useit.com/alertbox/participation_inequality.html , 2006. [47] "Online Communities and Their Impact on Bussiness: Ignore at Your Peril," Rubicon Consulting Inc., 2008. [48] Constance Elise Porter, "A Typology of Virtual Communities: A Multi- Disciplinary Foundation for Future Research," Journal of Computer-Mediated Communication, vol. 10, November 2004 2004. [49] Danah Boyd, "Teen Socialization Practices in Networked Publics," MacAthur Forum, Palo Alto, CaliforniaApril 23, 2008. [50] Rüdiger Schollmeier, "A Definition of Peer-to-Peer Networking for the Classification of Peer-to-Peer Architectures and Applications," in Proceedings of the First International Conference on Peer-to-Peer Computing, IEEE , 2002. [51] Salman A. Baset and Henning Schulzrinne, "An Analysis of the Skype Peer-to- Peer Internet Telephony Protocol," the Department of Computer Science, Columbia University 2004. [52] "SETI@home, " http://setiathome.ssl.berkeley.edu/ , 2005. [53] "BitTorrent, " http://www.bittorrent.com/ , 2010. [54] Stefan Saroiu, Krishna P. Gummadi, Richard J. Dunn, Steven D. Gribble and Henry M. Levy, "An Analysis of Internet Content Delivery Systems," in Proceedings of the 5th Symposium on Operating Systems Design and Implementation , 2002. [55] Mary Madden and Lee Rainie, "Music and video downloading moves beyond P2P " PEW INTERNET & AMERICAN LIFE PROJECT, March 2005. [56] "Gnutella, " http://g2.trillinux.org , 2010. [57] "eMule, " http://www.emule-project.net/ , 2010.

268

[58] S. Saroiu, K. Gummadi and S. Gribble, "A measurement study of peer-to-peer file sharing systems," in Proc. SPIE/ACM MMCN' 02 , San Jose, CA, 2002, pp. 156 - 170. [59] S. Sen and J. Wang, "Analyzing peer-to-peer traffic across large networks," in Proc. ACM IMW'02 Marseille, France, 2002, pp. 137 - 150. [60] "Kazaa, " http://www.kazaa.com/ , 2010. [61] S. Oyama, T. kokubo and T. Ishida, "Domain-Specific Web Search with Keyword Spices," IEEE TRANSACTION ON KNOWEDGE AND DATA ENGINEERING, vol. 16, pp. 17 - 27, 2004. [62] D. Butler, "Souped-Up Search Engines," Nature, vol. 405, pp. 112 - 115, 2000. [63] C. Shilakes and J. Tylman, Enterprise Information Portals : Merrill Lynch, 1998. [64] J. M. Firestone, "DKMS Brief No. Eight: Enterprise Information Portals And Enterprise Knowledge Portals," Executive Information Systems, Inc., 1999. [65] "CiteSeer, " http://citeseer.ist.psu.edu , 2010. [66] S. Lawrence, C. Giles and K. Bollacker, "Digital Libraries and Autonomous Citation Indexing," IEEE Computer, vol. 32, pp. 67 - 71, 1999. [67] A. McCallum, K. Nigam, J. Rennie and K. Seymore, "A Machine Learning Approach to Building Domain-Specific Search Engines," in Proc. 16th Int'l Jonit Conf. Artificial Intelligence (IJCAI-99) , 1999, pp. 662 - 667. [68] J. Qin, Z. Huang, Y. Zhou, M. Chau, C. Tseng, A. Yip, T.Ng, F. Guo, Z. Chen and H. Chen, "NanoPort: An Example for Building Knowledge Portals for Scientific Domains," in Proceedings. 2003 Joint Conference on Digital Libraries , 2003, p. 387. [69] R. C. Seacord, S. A. Hissam and K. C. Wallnau, "AGORA: A Search Engine for Software Components," IEEE INTERNET COMPUTING, vol. 2, pp. 62 - 70, 1998. [70] A. Kruger, C. Lee Giles, F. Coetzee, E. Glover, G. Flake, S. Lawrence and C. Omlin, "DEADLINER: Building a New Niche Search Engine," in Conference on Information and Knowledge Management Washington DC, 2000. [71] Y. Zhou, J. Qin, H. Chen, Z. Huang, Y. Zhang, W. Chung and G. Wang, "CMedPort: A Cross-regional Chinese Medical Portal," in Proceedings. 2003 Joint Conference on Digital Libraries , 2003, p. 379. [72] Erik Selberg and Oren Etzioni, "Multi-service search and comparison using the MetaCrawler," in Proceedings of the 4th World Wide Web Conference , 1995, pp. 195 - 208. [73] O. Etzioni, "Moving Up the Information Food Chain: Deploying Softbots on the World Wide Web," in Proc. 13th Nat'l Conf. Artificial Intelligence (AAAI-98) , 1998, pp. 509 - 516. [74] Fei-Yue Wang and Guanpi Lai, "Toward Application Specific Knowledge Engines," the Key Laboratory of Complex Systems and Intelligence Science, Chinese Academy of Sciences 2003. [75] Jon M. Kleinberg, "Authoritative sources in a hyperlinked environment," in Proceedings of ACM-SIAM Symposium on Discrete Algorithms , 1998, pp. 668 - 677. [76] S. Brin and L. Page, "The Anatomy of a Large-Scale Hypertextual Web Search Engine," Computer Networks and ISDN Systems, vol. 30, pp. 107-117, 1998.

269

[77] "Webopedia: Online Dictionary for Computer and Internet Terms and Definitions, " http://www.webopedia.com/ , 2010. [78] "Ajax (Programming), " http://en.wikipedia.org/wiki/Ajax_(programming) , 2010. [79] Shreeraj Shah, "Crawling Ajax-driven Web 2.0 Applications," 2007. [80] "Watir, " http://watir.com/ , 2010. [81] "Crowbar - SIMILE, " http://simile.mit.edu/wiki/Crowbar , 2010. [82] J. Cho, H. Garcia-Molina and L. Page, "Efficient Crawling through URL Ordering," in Proceedings of the 7th International World Wide Web Conference Brisbane, Australia, 1998. [83] O. Baujard, V. Baujard, S. Aurel, C. Boyer and R.D. Appel, "Trends in Medical Information Retrieval on the Internet," Computers in Biology and Medicine, vol. 28, pp. 589 - 601, 1998. [84] S. Chakrabarti, B. Dom and P. Indyk, "Enhanced Hypertext Categorization Using Hyperlink," in Proceedings of ACM SIGMOD International Conference on Management of Data Seattle, Washington, USA, 1998. [85] Du A-Ning, Bin-Xing Fang and Li Bin, "A Biased Support Vector Machine Approach to Web Filtering " Lecture Notes in Computer Science, vol. 3686, pp. 363 - 370, 2005. [86] M.F. Porter, "An algorithm for suffix stripping," Program, vol. 14, pp. 130 - 137, 1980. [87] Lawrence Philips, "The Double Metaphone Search Algorithm," C/C++ User Journal, 2000. [88] T. Berners-Lee, J. Hendler and O. Lassila, "The Semantic Web," Scientific American, vol. 284, pp. 34 -43, 2001. [89] "MindSwap, " http://www.mindswap.org/ , 2010. [90] "Knowledge Web, " http://knowledgeweb.semanticweb.org , 2010. [91] "MuseumFinland " http://www.museosuomi.fi , 2010. [92] "What is an Ontology?, " http://www-ksl.stanford.edu/kst/what-is-an- ontology.html , 1992. [93] Kristin M. Tolle and Hsinchun Chen, "Comparing noun phrasing techniques for use with medical digital library tools," Journal of the American Society for Information Science and Technology, vol. 51, pp. 352 - 370, 2000. [94] C. Fellbaum, WORDNET: An Electronic Lexical Database : MIT Press, 1998. [95] S. Decker, M. Erdmann, D. Fensel and R. Studer, "Ontobroker: Ontology based access to distributed and semi-structured information," in Proceedings of TC2/WG 2.6 8th Working Conference on Database Semantics (DS-8) , 1999. [96] R. Guha, R. McCool and E. Miller, "Semantic search," in Proceedings of WWW2003 , 2003. [97] Jeff Heflin and James Hendler, "Searching the web with shoe," in Proceedings of the AAAI Workshop on AI for Web Search , 2000, pp. 35 - 40. [98] K. Bartz, V. Murthi and S. Sebastian, "Logistic regression and collaborative filtering for sponsoreed search term recommendation," in Second Workshop on Sponsored Search Auctions , 2006. [99] J. S. Kandola, J. Shawe-Taylor and N. Cristianini, "Learning semantic similarity," in In NIPS : MIT Press, 2003, pp. 657 - 664.

270

[100] S. Greene, G. Marchionini, C. Plaisant and B. Shneiderman, "Previews and Overviews in Digital Libraries: Designing Surrogates to Support Visual Information Seeking," Journal of the American Society for Information Science, vol. 51, pp. 380-393, 2000. [101] P. Mclellan, A. Tombros, J. Jose, I. Ounis and M. Whitehead, "Evaluating Summarisation Technologies: A Task Oriented Approach," in Proceedings of the 3rd European Conference on Digital Libraries Paris, France, 2001, pp. 198 - 214. [102] D. McDonald and H. Chen, "Using Sentence Selection Heuristics to Rank Text Segments in TXTRACTOR," in Proceedings of the second ACM/IEEE-CS Joint Conference on Digital Libraries , Portland, Oregon, USA, 2002, pp. 28 - 35. [103] E. Hovy and C. Y. Lin, "Automated Text Summarization in SUMMARIST," in Advances in Automatic Text Summarization , I. Mani and M. T. Maybury, Eds.: MIT Press, 2003, pp. 81 - 94. [104] G. Salton, Automatic Text Processing: The Transformation, Analysis, and Retrieval of Information by Computer : Addison-Wesley, 1989. [105] T. Joachims, "Text Categorization with Support Vector Machines: Learning with Many Relevant Features," in the European Conference on Machine Learning , 1998. [106] H. Chen, H. Fan, M. Chau and D. Zeng, "MetaSpider: Meta-Searching and Categorization on the Web," Journal of the American Society for Information Science and Technology, vol. 52, pp. 1134 - 1147, 2001. [107] M. B. Amin and S. Shekhar, "Generalization by Neural Networks," in the 8th International Conference on Data Engineering , 1992. [108] L. Baker and A. McCallum, "Distributional Clustering of Words for Text Classification," in SIGIR-98 , 1998. [109] D. S. D. Michie and C. Taylor, Machine Learning, Neural and Statistical Classification : Ellis Horwood, 1994. [110] Eui-Hong (Sam) Han, George Karypis and Vipin Kumar, "Text Categorization Using Weight Adjusted k-Nearest Neighbor Classification," in Proceedings of the 5th Pacific-Asia Conference on Knowledge Discovery and Data Mining , 1999, pp. 53 - 65. [111] W. Lam and C. Y. Ho, "Using A Generalized Instance Set for Automatic Text Categorization," in SIGIR-98 , 1998. [112] B. Larsen and C. Aone, "Fast and Effective Text Mining Using Linear-time Document Clustering," in the 5th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , 1999, pp. 16 - 22. [113] O. Zamir and O. Etzioni, "Grouper: A Dynamic Clustering Interface to Web Search Result," in the Eighth World Wide Web Conference Toronto, Canada, 1999, pp. 1361 - 1374. [114] T. Ong and H. Chen, "Updatable PAT-Tree Approach to Chinese Key Phrase Extraction Using Mutual Information: A Linguistic Foundation for Knowledge Management," in the Second Asian Digital Library Conference Taipei, Taiwan, 1999, pp. 63 - 84. [115] S. K. Card, J. D. Mackinlay and B. Shneiderman, Readings in Information Visualization: Using Vision to Think . San Francisco, CA: Morgan Kaufmann, 1999.

271

[116] C. Lin, H. Chen and J. F. Nunamaker, "Verifying the Proximity Hypothesis for Self-Organizing Maps," Journal of Management Information Systems, vol. 16, pp. 57 - 70, 2000. [117] W. Chung, T. Elhourani, A. Bonilla, G. Lai, W. Xi and H. Chen, "Supporting Information Seeking Multnational Organizations: A Knowledge Portal Approach," in Proceedings of the 38th HAWAII INTERNATIONAL CONFERENCE ON SYSTEM SCIENCES (HICSS'05) , Hawaii, USA, 2005. [118] C. Ahlberg and B. Shneiderman, "Visual Information Seeking: Tight Coupling of Dynamic Query Filters with Starfield Displays," in ACM CHI' 94, 1994, pp. 313 - 317. [119] A. Spoerri, "InfoCrystal: A Visual Tool for Information Retrieval," in IEEE Visualization 93 , 1993, pp. 150 - 157. [120] T. Kohonen, "The Self-Organizing Map," Proceedings of The IEEE, vol. 78, pp. 1464 - 1480, 1990. [121] K. Hatano, Q. Qian and K. Tanaka, "A SOM-Based Information Organizer for Text and Video Data," in Proceedings of The 5th International Conference on Database Systems for Advanced Applications (DASFAA’97) , 1997, pp. 205 - 214. [122] S. Cutter and T. J. Wilbank, The Geographical Dimensions of Terrorism : Taylor & Francis, Inc., 2003. [123] L. W. Kennedy and C. M. Lunn, "Developing a Foundation for Policy Relevant Terrorism Research in Criminology," Rutgers University, New Brunswick. [124] G. Weimann, " www.terror.net: How Modern Terrorism Uses the Internet," United States Institute of Peace, 2004. [125] E. O. F. Reid, "An Analysis of Terrorism Literature: A Bibliometric and Content Analysis Study," School of Library and Information Management, University of Southern California, Los Angels1983. [126] "Making the Nation Safer: the Role of Science and Technology in Countering Terrorism," the National Research Council, Washington, D. C.2002. [127] Gail M. Hodge, "Best Practices for Digital Archiving: An Information Life Cycle Approach," D-Lib Magazine, vol. 6, 2000. [128] "About the Terrorism Research Center," Terrorism Research Center, 2003. [129] "Web Portal and Portal Taxonomies," Carnegie Mellon University, 2001. [130] E. Reid, J. Qin, W. Chung, J. Xu, Y. Zhou, R. Schumaker, M. Sageman and H. Chen, "Terrorism Knowledge Discovery Project: A Knowledge Discovery Approach to Addressing the Threats of Terrorism," in Proceedings of the Second Symposium on Intelligence and Security Informatics , Tucson, AZ, 2004, pp. 125 - 145. [131] J. Carbonell and J. Goldstein, "The Use of MMR: Diversity-based Reranking for Reordering Documents and Producing Summaries," in the 21st Annual International ACM-SIGIR Conference on Research and Development in Information Retrieval Melbourne, Australia, 1998. [132] M. A. Hearst, "Multi-paragraph Segmentation of Expository Text," in the 32nd Annual Meeting of the Association for Computational Linguistics Las Cruces, New Mexico, 1994. [133] M. Chau, H. Chen, J. Qin, Y. Zhou, W.-K. Sung, Y. Chen, Y. Qin, D. McDonald, A. Lally and M. Landon, "NanoPort: A Web Portal for Nanoscale Science and

272

Technology," in the 2nd ACM/IEEE-CS Joint Conference on Digital Libraries (JCDL'02) Portland, OR, USA, 2002. [134] H. Chen and K. J. Lynch, "Automatic Construction of Networks of Concepts Characterizing Document Databases," IEEE Transactions on Systems, Man, and Cybernetics, vol. 22, pp. 885 - 902, 1992. [135] H. Chen, A. Lally, B. Zhu and M. Chau, "HelpfulMed: Intelligent Searching for Medical Information over the Internet," Journal of the American Society for Information Science and Technology, vol. 54, pp. 683 - 694, 2003. [136] "Movie piracy costs Hollywood 75 pct more than feared : report," in AFP , 2006. [137] Jorma Mellin, "Peer-to-Peer Networking - Phenomenon and impacts to carriers," in Telecom Forum , 2004. [138] "MP3, " http://en.wikipedia.org/wiki/MP3 , 2010. [139] "WINAMP, " http://www.winamp.com , 2010. [140] "DivX - Play DivX, AVI & MKV videos on computers and devices, " http://www.divx.com , 2010. [141] "Xvid.org: Home of the Xvid Codec, " http://www.xvid.org/ , 2010. [142] "Napster, " http://www.napster.com , 2010. [143] "eDonkey2000, " http://www.edonkey2000.com/ , 2005. [144] "MLdonkey, " http://mldonkey.sourceforge.net , 2010. [145] Xiaohe Lu, "On P2P File-Sharing: A Major Problem: A Chinese Perspective," Journal of Business Ethics, vol. 63, pp. 63 - 73, 2006. [146] W. D. Walls, "Cross-country analysis of movie piracy," Applied Economics, vol. 40, pp. 625 - 632, 2008. [147] Birgitte Andersen and Marion Frenz, "The Impact of Music Downloads and P2P File-Sharing on the Purchase of Music: A Study for Industry Canada," Industry Canada, 2008. [148] I. Clarke, T. W. Hong, Scott G. Miller, O. Sandberg and B. Wiley, "Protecting free expression online with freenet," IEEE Internet Computing, vol. 6, pp. 40 - 49, 2002. [149] "Mnet, " http://mnetproject.org/ , 2005. [150] R. Bhagwan, S. Savage and G. Voelker, "Understanding availability," in Proceedings of the 2nd International Workshop on Peer-to-Peer Systems , 2003. [151] Y. Qiao and F.E. Bustamante, "Structured and Unstructured Overlays Under the Microscope - A Measurement based View of Two P2P Systems That People Use," in Proceedings of the USENIX Annual Technical Conference , 2006.