Themen Des Information Retrieval

Total Page:16

File Type:pdf, Size:1020Kb

Themen Des Information Retrieval Andreas Henrich Daniel Blank (Hrsg.) Themen des Information Retrieval Suchmaschinen und Web-Suche Beiträge des Seminars im Sommersemester 2012 Lehrstuhl für Medieninformatik Otto-Friedrich-Universität Bamberg Inhaltsverzeichnis Designing Search User Interfaces The Design of Search User Interfaces 5 Eduard Anton Models of the Information Seeking Process 21 Andreas Hünlein Search User Interfaces: Presentation of Search Results 35 Alexander Schreiner Query specification in information retrieval 49 Alexander Knogl Popular Destinations in Kombination mit anderen 69 Methoden der Unterstützung von Query Reformulation Felix Wiedemann Information Visualization for Search Interfaces 85 Dietlinde Flavia Lump Information Visualization for Text Analysis 105 Johannes Rettig A Survey of Personalization Techniques for Online Search 127 and Recommender Systems Ralf Strobel Evaluierung im Information Retrieval Die Evaluierungsinitiative INEX 145 Christoph Jungnickl Techniken zur Identifikation typischer Wikipedia Anfragen 163 Florian Gundelsheimer Software-Bibliotheken für das Information Retrieval Sphinx - Open search server 179 Manuel Leithner Apache Solr 199 Johannes Döring MapReduce mit Apache Hadoop 217 Philipp Ott Aspekte des Enterprise Search Exploring Query Patterns in Email Search 233 Daniel Thomä Nutzungsrechte in Enterprise Search Systemen 249 - Secure Enterprise Search Michael Großmann 5 Seminararbeiten The Design of Search User Interfaces Eduard Anton Otto-Friedrich-Universität-Bamberg, Lehrstuhl für Medieninformatik [email protected] Zusammenfassung. Eine Suchmaschine muss eine enorme Anzahl von Use Cases, über eine enorme Ansammlung von Informationen, benutzt von einem weiten Feld von Menschen unterstützen und muss dabei ein hohes Maß an Gebrauchstauglichkeit unterstützen. Die Gebrauchstauglichkeit und das Interface Design gehen dabei Hand in Hand. Die Wissenschaft, die sich damit befasst, nennt man Mensch-Computer-Interaktion. Ziel dieser Arbeit ist es, die Schwierigkeiten von Suchmaschinennutzern mit dem Interface Design von Suchmaschinen aufzuzeigen und die von der Mensch-Computer-Interaktion entwickelten Richtlinien für ein erfolgreiches Interface Design vorzustellen, um diese Schwierigkeiten zu lösen und an Beispielen in der Praxis zu zeigen. Ausgangspunkt dieser Arbeit stellt das erste Kapitel vom Buch Search User Interfaces von Marti A. Hearst [1] dar. Schlagworte: Mensch-Computer-Interaktion, Gebrauchstauglichkeit, Design Richtlinien 1 Einleitung Berühmte Persönlichkeiten, Preisvergleiche oder Informationen für ein bestimmtes Hausarbeitsthema - in der heutigen Gesellschaft ist die Informationsbeschaffung ein wichtiges Thema. Wo man früher in Bibliotheken, Büchereien und Zeitungsarchiven suchte, recherchiert man heute mit Suchmaschinen im Internet. Millionen von Menschen nutzen dabei täglich unterschiedliche Suchmaschinen. Um die Informationen zugänglich und nutzbar zu machen, braucht man effektive und effiziente Suchmaschinen. Dabei spielt das Design von Suchinterfaces eine wichtige Rolle, es „verwandelt unstrukturierte Rohdaten in brauchbare Informationen und beschreibt die Form der dargestellten Informationen. Es hat Einfluss darauf, wie die Nutzer weitere Informationen verarbeiten sowie rezipieren und trägt dazu bei, ob das Produkt wieder verwendet wird. Bei einem positiven Erlebnis kommen die Nutzer gerne wieder“ [2, S. 228]. Die Nutzung eines Suchinterfaces bringt aber auch Probleme mit sich. Vor allem unerfahrene User, ältere Menschen und Kinder haben Schwierigkeiten bei ihrer Informationssuche im Internet. Diese Schwierigkeiten führen darauf zurück, dass Menschen bei ihrer Informationssuche, bevor die Verfügbarkeit von Computern und Sommersemester 2012 Seminararbeiten 6 Internet so weit verbreitet war wie sie heute ist, sich andere Menschen zu Nutze machten. So fragte man Bibliothekare, wo eine bestimmte Information zu finden sei. Diese verstanden, was gesucht wird, obwohl bei der Frage vielleicht nicht die richtige Syntax oder richtigen Begriffe genannt wurden. Jetzt müssen Leute mit Suchmaschinen interagieren, obwohl sie sich mit moderner Informationsbeschaffung nur wenig oder gar nicht auskennen(vgl. [3, S. 1-2]). Um diese Interaktion für alle zu ermöglichen, ist es wichtig, das Interface möglichst zu simplifizieren. Dies wird im nächsten Kapitel näher erläutert. 2 Das Interface einfach halten Nahezu bei jeder Suchmaschine tippt man ein Stichwort in ein Sucheingabefeld ein und die Ergebnisse werden in einer vertikalen Liste ausgegeben. Dabei hat sich auch in den letzten Jahren nicht viel verändert. Vergleicht man eine Informationssuche, mit dem gleichen Suchbegriff von 1997,2007 und heute, stellt man fest, dass keine größeren, gravierenden Veränderungen im Interface bestehen und es relativ einfach gehalten wird (siehe Abbildung 1). Abbildung 1. Ergebnislisten (links von 1997, rechts von 2007) einer Googlesuche mit dem Suchbegriff „darter habitat“ [1] Warum verändert sich das Standardinterface kaum? Gründe hierfür liegen beim Zweck der Suche. Bei einer Suche möchte der Nutzer seinem Informationsbedürfnis nachgehen. Dabei wirkt ein überladenes und aufdringliches Interface störend. Außerdem lenkt ein solches Interface beim Lesen ab, was ohnehin eine mental sehr Lehrstuhl für Medieninformatik, Otto-Friedrich-Universität Bamberg 7 Seminararbeiten intensive Tätigkeit ist, die kaum multitaskfähig ist, und deshalb die volle Aufmerksamkeit des Nutzers erfordert. Außerdem soll die Simplizität des Interfaces auch für die bereits angesprochenen Menschen, die Problemen bei der Bedienung mit Suchmaschinen haben, verständlich sein. Trotz des ohnehin einfach gehaltenen Interfaces haben Studien gezeigt, dass eine weitere Vereinfachung viele gemachte Fehler reduzieren würde. Fehler sind beispielweise die falsch angewandte Syntax in der Suchzeile, da Suchzeile und Adresszeile miteinander verwechselt werden, die Missinterpretation der Boolean Operatoren und die falsche Verwendung der Begriffreihenfolge in der Abfrage. In einem Versuch, die Websuche für ältere Leute zu vereinfachen, wurde ein Suchinterface gestaltet, das viele gebräuchliche Fehler reduzieren soll. Dieses Interface hat eine größere Suchleiste für die Abfragen und mehr Platz zwischen den Buchstaben, um es einfacher zu machen, Abfragen zu bearbeiten. Außerdem wurde ein Button hinzugefügt, der die Suchleiste leert, um neue Abfragen schneller starten zu können. Andere Vereinfachungen an der Ergebnisliste wurden zusätzlich vorgenommen, um die Resultate besser zu verstehen. Viele Nutzer dieses Versuchs verstanden die Funktionalität und schätzten die Klarheit (vgl. [3, S. 62-63]). Man sieht also, dass eine weitere Vereinfachung durchaus möglich und sinnvoll ist, was auch das Scheitern mancher Versuche, komplexere Interfaces einzuführen, erklärt. 3 Suchinterfaces vor und nach dem Web Modernes Information Retrieval, wie wir es heute haben, hat „vom Memex über den Sputnikschock zum Weinberg-Report" [8, S. 38] einen langen Weg hinter sich. Bevor das Internet ein weltweites Phänomen wurde, mussten die Menschen zum Lexikon greifen oder in anderen Büchern nachschlagen, um an ihre gewünschten Informationen zu gelangen. Die computergestützte Informationsbeschaffung war üblicherweise ein Privileg von ausgebildeten Nutzern wie Rechtsanwaltfachangestellten, Journalisten und Bibliothekare. Sie suchten in bibliografischen Datensätzen, Rechtsfällen oder Nachrichtensammlungen. Normalerweise wurde offline nach dem Ort der Quelle gesucht und anschließend wurde eine Papierkopie gemacht. Heute findet man sämtliche Texte direkt im Internet und der Text ist somit sofort verfügbar und lesbar. Ein weiterer Unterschied besteht darin, dass alte Systeme, bevor Grafikdisplays üblich waren, auf Kommandozeilen- Interfaces basiert haben. Das hatte zufolge, dass komplexe Kombinationen von Operatoren gemerkt werden mussten und ein Verständnis für Boolean Operatoren mit sich gebracht werden musste. Das Verständnis für Boolean Operatoren und Kommandosprache war nur einer kleinen Anzahl von Leuten bekannt und so gab es nur wenige, die Zugang zu den Inhalten hatten. Ein letzter Unterschied besteht darin, dass damals Suchmaschinen entweder nach Zeit, Anzahl der Abfragen oder Anzahl der Ergebnisse, Geld gekostet haben. Heute sind Suchmaschinen durch den großen Wettbewerb meistens kostenlos. Sommersemester 2012 Seminararbeiten 8 Diese Gegensätze helfen dabei, die Unterschiede der Suchinterfaces vor und nach dem Web zu verstehen. 4 User Interfaces in der Mensch-Computer Interaktion Die Mensch-Computer Interaktion (Human-Computer Interaction) ist die Wissenschaft, die sich mit dem Interface Design und der Kommunikation zwischen Mensch und Computern auseinandersetzt. Das UCD (User Centered Design) ist eine Entwicklung der Mensch-Computer Interaktion. Es ist ein „ Vorgehensmodell zur benutzerzentrierten Gestaltung interaktiver Systeme“ [2, S. 226-227]. Das bedeutet, bei der Entwicklung einer Suchmaschine wird der zukünftige Nutzer mit seinen Aufgaben, Zielen und Eigenschaften in den Mittelpunkt des Entwicklungsprozesses gestellt. Dies wird durch ein iteratives Vorgehen über mehrere Phasen erreicht. Dadurch erhofft man sich die Entwicklung einer Suchmaschine, die über eine hohe Gebrauchstauglichkeit (Usability) verfügt. Die Gebrauchstauglichkeit ist eine qualitative Eigenschaft eines User Interface und mit dem Konzept der Benutzerfreundlichkeit vergleichbar. Die Gebrauchstauglichkeit besitzt bestimmte Attribute, die lauten: Erlernbarkeit: wie schnell und einfach ein Nutzer in einem neuen System
Recommended publications
  • Statistics for Donauschwaben-Usa.Org (2009-03)
    Statistics for donauschwaben-usa.org (2009-03) Statistics for: donauschwaben-usa.org Last Update: 03 Apr 2009 - 14:14 Reported period: Month Mar 2009 When: Monthly history Days of month Days of week Hours Who: Organizations Countries Full list Hosts Full list Last visit Unresolved IP Address Robots/Spiders visitors Full list Last visit Navigation: Visits duration File type Viewed Full list Entry Exit Operating Systems Versions Unknown Browsers Versions Unknown Referrers: Origin Referring search engines Referring sites Search Search Keyphrases Search Keywords Others: Miscellaneous HTTP Status codes Pages not found Summary Reported period Month Mar 2009 First visit 01 Mar 2009 - 00:17 Last visit 31 Mar 2009 - 23:17 Unique visitors Number of visits Pages Hits Bandwidth 2112 2781 15381 71620 4.59 GB Viewed traffic * (1.31 visits/visitor) (5.53 Pages/Visit) (25.75 Hits/Visit) (1732.12 KB/Visit) Not viewed traffic * 8539 10927 896.73 MB * Not viewed traffic includes traffic generated by robots, worms, or replies with special HTTP status codes. Monthly history Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 Month Unique visitors Number of visits Pages Hits Bandwidth Jan 2009 0 0 0 0 0 Feb 2009 0 0 0 0 0 Mar 2009 2112 2781 15381 71620 4.59 GB Apr 2009 0 0 0 0 0 May 2009 0 0 0 0 0 Jun 2009 0 0 0 0 0 Jul 2009 0 0 0 0 0 Aug 2009 0 0 0 0 0 Sep 2009 0 0 0 0 0 Oct 2009 0 0 0 0 0 Nov 2009 0 0 0 0 0 Dec 2009 0 0 0 0 0 Total 2112 2781 15381 71620 4.59 GB Days of month 01 02 03 04 05 06 07 08
    [Show full text]
  • 2008 New Media M&A Round-Up
    2008 New Media M&A Round‐Up The Year in Digital Media Mergers, Acquisitions & Capital Raises PEACHTREE MEDIA ADVISORS, INC. N EW M EDIA I NVESTMENT B ANKING EACHTREE EDIA DVISORS NC P M A , I . Better Service ▪ Lower Fees New Media Mergers & Acquisitions TABLE OF CONTENTS I. Internet/New Media M&A Transactions by Sector 1 II. M&A Transactions & Capital Raised in 2008 12 III. 2008 Interactive Media Valuations/Comps 31 IV. Conclusion/2009 Outlook 37 V. Out‐of‐Home/Alternative M&A Transactions 38 VI. Conclusion/2009 Outlook 38 VII. 2008 OOH Valuations/Comps 41 Peachtree Media Advisors, Inc. Peachtree Media Advisors, Inc. is a New York based investment bank serving the out‐of‐ home and interactive marketing sectors of media. The company provides mergers, acquisitions and capital raise advisory services to lower middle‐market companies in the two fastest growing sectors of media. John Doyle, Managing Director & Founder, has been a media investment banker for more than 12 years; closed and structured more than 22 deals; and has a strong knowledge‐base of financial and strategic buyers in these sectors. If you are interested in learning more about valuation, positioning, preparation or the merger and acquisition process, please go to www.PeachtreeMediaAdvisors.com or contact John Doyle at (212) 570‐1009. John H. Doyle II Managing Director & Founder Peachtree Media Advisors, Inc. 50 Vanderbilt Ave., #30 New York, NY 10017 PH. 212.570.1009 ▪ FAX 646.607.1786 www.peachtreemediaadvisors.com Table of Contents Better Service ▪ Lower Fees New Media Mergers & Acquisitions Online Media M&A Activity in 2008 Although the Enabling, Analytics and Ad Serving category had the fourth highest In 2008, there were 707 merger, acquisition level of reported transaction value in 2008, and capital raise transactions in the online this category had the largest percentage sector of media (92 more transactions than increase in capital flowing to it than any the 615 in 2007).
    [Show full text]
  • January 20, 2007 the Free-Content News Source That You Can Write! Page 1
    January 20, 2007 The free-content news source that you can write! Page 1 Top Stories Wikipedia Current Events Regional High School in Sudbury has left a 15-year old student Turkish-Armenian journalist the attack. dead. killed in Turkey •A close aide to UK Prime Minister An Armenian Tony Blair is arrested in a The stabbing happened around journalist, Hrant corruption probe. 7:20 am EST, before classes had Dink, was started. A fight broke out in a assassinated •Hrant Dink, an Armenian-Turkish boy's bathroom between the 15- today at the age writer, is shot dead in Istanbul. year old victim, James Alenson and of 52 in front of Dink was convicted in 2005 for 16-year-old suspect John Odgren, the Agos 'insulting Turkishness' by writing the fight spilled out in the hallway, newspaper office, an article on the Armenian where the stabbing occured. at the Istanbul district of Genocide. Osmanbey, where he worked as •Hurricane force winds claim at The school was sent into a the editor and a journalist. least 40 lives in Western Europe "lockdown" and students were including 10 lives in Britain and ushered into the gym, cafeteria H5N1 Avian Flu virus mutates; 11 in Germany, and other victims and various classrooms. Alenson shows resistance to drugs in The Netherlands (6), Poland was rushed to Emerson Hospital A mutation in the H5N1 Avian Flu (6), Czech Republic (3) and where he was pronounced dead at or Bird Flu France (3). 8:15 am EST. Odgren admitted to virus has the stabbing and was in the •Robert Gates, the United States been found principal's office saying "I did it, I Secretary of Defense, visits in two did it," to police.
    [Show full text]
  • Download Download
    International Journal of Management & Information Systems – Fourth Quarter 2011 Volume 15, Number 4 History Of Search Engines Tom Seymour, Minot State University, USA Dean Frantsvog, Minot State University, USA Satheesh Kumar, Minot State University, USA ABSTRACT As the number of sites on the Web increased in the mid-to-late 90s, search engines started appearing to help people find information quickly. Search engines developed business models to finance their services, such as pay per click programs offered by Open Text in 1996 and then Goto.com in 1998. Goto.com later changed its name to Overture in 2001, and was purchased by Yahoo! in 2003, and now offers paid search opportunities for advertisers through Yahoo! Search Marketing. Google also began to offer advertisements on search results pages in 2000 through the Google Ad Words program. By 2007, pay-per-click programs proved to be primary money-makers for search engines. In a market dominated by Google, in 2009 Yahoo! and Microsoft announced the intention to forge an alliance. The Yahoo! & Microsoft Search Alliance eventually received approval from regulators in the US and Europe in February 2010. Search engine optimization consultants expanded their offerings to help businesses learn about and use the advertising opportunities offered by search engines, and new agencies focusing primarily upon marketing and advertising through search engines emerged. The term "Search Engine Marketing" was proposed by Danny Sullivan in 2001 to cover the spectrum of activities involved in performing SEO, managing paid listings at the search engines, submitting sites to directories, and developing online marketing strategies for businesses, organizations, and individuals.
    [Show full text]
  • Web Search Engine
    Web Search Engine Bosubabu Sambana, MCA, M.Tech Assistant Professor, Dept of Computer Science & Engineering, Simhadhri Engineering College, Visakhapatnam, AP-531001, India. Abstract: This hypertext pool is dynamically changing due to this reason it is more difficult to find useful The World Wide Web (WWW) allows people to share information.In 1995, when the number of “usefully information or data from the large database searchable” Web pages was a few tens of millions, it repositories globally. We need to search the was widely believed that “indexing the whole of the information with specialized tools known generically Web” was already impractical or would soon become as search engines. There are many search engines so due to its exponential growth. A little more than a available today, where retrieving meaningful decade later, the GYM search engines—Google, information is difficult. However to overcome this problem of retrieving meaningful information Yahoo!, and Microsoft—are indexing almost a intelligently in common search engines, semantic thousand times as much data and between them web technologies are playing a major role. In this providing reliable sub second responses to around a paper we present a different implementation of billion queries a day in a plethora of languages. If this semantic search engine and the role of semantic were not enough, the major engines now provide much relatedness to provide relevant results. The concept higher quality answers. of Semantic Relatedness is connected with Wordnet which is a lexical database of words. We also For most searchers, these engines do a better job of made use of TF-IDF algorithm to calculate word ranking and presenting results, respond more quickly frequency in each and every webpage and to changes in interesting content, and more effectively Keyword Extraction in order to extract only useful eliminate dead links, duplicate pages, and off-topic keywords from a huge set of words.
    [Show full text]
  • TX−Hidalgocounty
    TX−HidalgoCounty Web Log Analysis Custom Date Range Report Report Range:06/01/2008 00:00:00 − 01/09/2009 23:59:59 This report was generated by WebTrends(R) Monday January 12, 2009 − 09:47:21 Final report conversion by WebTrends Document Utility, Version 6.1a (build 419) (c) 1996−2003 NetIQ Corporation. All rights reserved. Table of Contents Overview Dashboard....................................................................................................................................................1 Commerce Dashboard..................................................................................................................................................3 Marketing Dashboard...................................................................................................................................................5 Visitors Dashboard.......................................................................................................................................................7 Pages Dashboard.........................................................................................................................................................11 Navigation Dashboard................................................................................................................................................13 Technical Dashboard..................................................................................................................................................15 Activity Dashboard.....................................................................................................................................................17
    [Show full text]
  • Index of /Sites/Default/Al Direct/2009/May
    AL Direct, May 6, 2009 Contents U.S. & World News ALA News AL Focus Booklist Online Chicago Update Division News Awards Seen Online Tech Talk The e-newsletter of the American Library Association | May 6, 2009 Publishing Actions & Answers Calendar U.S. & World News Guantanamo records to be preserved New York University’s Tamiment Library and Seton Hall University’s Center for Policy and Research have announced a project to document, preserve, and provide access to legal records and other documents of the Guantanamo Bay Detention Camp. The materials will include lawyers’ files, oral histories of the attorneys and detainees, Department of Defense websites, photographs and videotapes, and electronic records, as well as records relating to the rules governing enemy combatants, prisoner interrogation, and the government’s representation of battlefield capture.... American Libraries Online, May 6 Dismissal of Wisconsin board members draws national censure ALA has joined publishing and free-speech groups in condemning the West Bend, Wisconsin, common council for not reappointing four library board members after they failed to act on a citizens group’s call to restrict the availability of sexually explicit books. At its April 21 meeting, the council voted 5–3 to reject Mayor Kristine Deiss’s recommendation to reappoint the four library board members whose terms were concluding: Mary Reilly-Kliss, Tom Fitz, James Pouros, and Alderman Nick Dobberstein.... ALA Annual Conference, American Libraries Online, May 4 Chicago, July 9–15. The 2009 ALA Annual Poster Lexington Director Imhoff: Scrutiny of finances Sessions will be held July inaccurate 11–13 at McCormick Place “There is a situation going on in Lexington [Kentucky],” said Kathleen West.
    [Show full text]
  • Regulating Search
    Harvard Journal of Law & Technology Volume 22, Number 2 Spring 2009 REGULATING SEARCH Viva R. Moffat* TABLE OF CONTENTS I. INTRODUCTION..............................................................................475 II. SEARCH ENGINES AND SEARCH ENGINE DISPUTES......................479 A. The Way Search Engines Search and Store Content................482 B. The Way Search Engines Display Content...............................483 C. The Way Search Engines Make Money....................................484 III. TRACING THE SCHOLARLY DEBATE ON SEARCH ENGINE REGULATION.................................................................................487 A. The Case for Agency Regulation..............................................487 B. The Case for Market Regulation ..............................................490 C. Some Problems with the Solutions at Either End of the Spectrum ................................................................................491 1. Concerns About Agency Regulation.....................................492 2. Concerns About the Free Market Approach..........................495 IV. AN ALTERNATIVE IN THE BIPOLAR DEBATE: A FEDERAL FORUM FOR SEARCH ENGINE DISPUTES........................................498 A. A Federal Forum Compared to More Centralized Regulation..............................................................................500 1. Flexibility Allows the Common Law to Accommodate Changing Technology.....................................................500 2. The Flexibility of the Common Law Renders
    [Show full text]
  • Radio Frequency Identification Based Smart
    ISSN (Online) 2394-2320 International Journal of Engineering Research in Computer Science and Engineering (IJERCSE) Vol 3, Issue 10, October 2016 History and Web Search Engines Works [1]Saurabh Kumar, [2] Maneesh Kumar, [3] Vipul Agarwal M .Tech Scholar, Department of Computer Science and Engineering, Dr. A.P.J Abdul Kalam Technical University, Lucknow (U.P), Abstract: -- A web search engine is a software system that is designed to search for information on the World Wide Web. The search results are generally presented in a line of results often referred to as search engine results pages(SERPs). The information may be a mix of web pages, images, and other types of files. Some search engines also mine data available in databases or open directories. Unlike web directories, which are maintained only by human editors, search engines also maintain real- time information by running an algorithm on a web crawler. I. INTRODUCTION Lycos Active Search engine is a web software program or Infoseek Inactive web based script available over the Internet that searches documents and files for keywords and returns the list of results containing those keywords. Today, there are 1995 AltaVista Inactive, redirected to Yahoo! numbers of different search engines available on the Internet, each with their own techniques and specialties. Daum Active Search Engine Optimization is a technique to improve visibility of a website in search engine. Magellan Inactive II. HISTORY Excite Active Timeline (full list) SAPO Active Year Engine Current status Yahoo! Active,
    [Show full text]
  • Building a Web-Scale Search Engine with Perl
    Building a Web-Scale Search Engine with Perl Greg Lindahl, CTO, blekko [email protected] - @glindahl - wumpus" Waah •# This is my first YAPC •# They scheduled me against Rick?! Agenda •# Personal histories •# The Search Business in 3 minutes •# Why we chose Perl •# WriAng NoSQL in Perl •# The search engine app – event-driven progr. •# ConAnuous Everything •# Updang perl (and CentOS) •# Open Data / Open Source •# Our secret sauce Personal Histories •# me: icon, usenet? or irc?, merlyn, camel book •# hp://www.pbm.com/~lindahl/ •# really a supercompuAng guy: Fortran, MPI •# Contrib to ircd (founded efnet), binuls, emacs •# others: GnuHoo => NewHoo => ODP / dmoz •# Netscape => AOL => AOL-TimeWarnerMegaCorp •# Popular open data dataset, inspired Wikipedia •# All of us know a lot of other languages The Search Business in 3 minutes •# Some people say building a real search engine costs $1 billion –#and by real, I mean mulA-billion-page crawl and index, not using Google/bing’s index •# Recent “real” failures: cuil, SearchMe; they raised ~ $30-40 million, hire 80-100 people •# The only successful new “real” engine since Google is the bing re-write! ($1B loss/year) •# We felt real innovaon was possible only if we had our own crawl and index What our marketing team says we built" API Partners Search Technology Distributed CompuAng Plaorm Crawler Indexer Machine Learning Ranker Query classifier Content API’s Ad API’s Crawled Web Our plan •# Build the basics of a search engine •# Try some innovaons, most of which will fail •# Don’t die on launch day, have a long runway •# Try some innovaons, most of which will fail •# … •# Profit! •# (we didn’t really have a plan) •# keep the team small: build an environment that makes programmers efficient Choosing perl •# “Maybe we should switch to Python –#basically the same language –#easier to hire” •# Went and bought Python books •# 1 week later: “Anyone read their book?” •# Allrighty, Perl it is.
    [Show full text]
  • Search Engine: a Survey
    [Singh , 2(1): Jan., 2013] ISSN: 2277 -9655 IJESRT INTERNATIONAL JOURNA L OF ENGINEERING SCI ENCES & RESEARCH TECHNOLOGY Search Engine : A Survey Ankur Singh Bist Govind Ballabh P ant University of Agriculture And Technology, Pant, India [email protected] Abstract There is lot of information available on the internet so to fetch the relevant information is very typical task . In this paper our purpose is to make a survey on various issues related to search engines li ke semantic web search technology. Keywords: Crawler, Ontology .. Introduction AlltheWeb Inactive (URL redirected to Yahoo!) Searching is an important issue to retrieve the GenieKnows Active, rebranded Yellowee.com required data . Search engines are serving the same 1999 Naver Active task . There are various challenges covered by search Teoma Active engines in recent years . There are lot of challenges Viv isimo Inactive still to be solved . Answer the intelligent questions Baidu Active most efficiently is one of the major issue that is to be 2000 handled by search engines semantic web searching Exalead Inactive approaches are evolving that are responsible for 2002 Inktomi Acquired by Yahoo! making an appropriate results for intelligent input 2003 Info.com Active query . There are also various issues related to Active, Launched own web search Yahoo! Search vertical search engines that needed to be handled by (see Yahoo! Directory, 1995) 2004 applying efficient approaches .The timeline of search A9.com Inactive engines is provided below:-- Sogou Active AOL Search Active
    [Show full text]
  • [Ccebook.Cn]The Economist October 25Th 2008
    www.EliteBook.net Search Welcome Economist.com My accountManage my newsletters Log out Requires subscription Friday October 24th 2008 Site feedback Home Print edition October 25th 2008 This week's print edition Into the storm Previous print editions Subscribe Daily news analysis How the emerging world copes Oct 18th 2008 Subscribe to the print edition Opinion with the tempest will affect the Oct 11th 2008 All opinion Or buy a Web subscription for world economy and politics for Oct 4th 2008 full access online Leaders a long time: leader Sep 27th 2008 Letters to the Editor Sep 20th 2008 RSS feeds Blogs Receive this page by RSS feed Columns More print editions and KAL's cartoons covers » Correspondent's diary Economist debates The world this week World politics Politics this week All world politics Business this week Politics this week KAL's cartoon International A special report on corporate IT United States The Americas Leaders Let it rise Asia Middle East and Africa The financial crisis Where the cloud meets the ground Europe Into the storm Britain Creating the cumulus Land reform in China Special reports Still not to the tiller On the periphery Business All business Information technology Highs and lows Business this week Clouds and judgment Management The long nimbus Iraq Business education When to call the soldiers home Computers without borders Finance and economics The state as owner All finance and economics Sources and acknowledgments Economics focus Re-bonjour, Monsieur Colbert Economics A-Z Offer to readers Letters Markets and
    [Show full text]