SCHOOL OF COMPUTER SCIENCE
Semantic Wiki for Travel and Holidays using OWL
Project Initial Report COMP60990
Student Name: Joao Rafael Goncalves Student Number: 7387762
Supervisor(s): Prof. Alan Rector Dr. Robert Stevens Date: 27 Apr 2009 Table Of Contents
Table Of Contents...... 2 1. Abstract...... 3 2. Introduction...... 4 3. Background and related literature...... 6 3.1 Semantic Wiki approaches ...... 6 3.2 Wiki Architectures...... 7 3.2.1 General structure of a Wiki ...... 7 3.2.2 General structure of a Semantic Wiki...... 8 3.3 Types of Semantic Wikis...... 8 3.4 Semantic Wiki implementations...... 9 3.4.1 Features Overview...... 9 3.4.2 Semantic MediaWiki ...... 10 3.4.3 BOWiki...... 11 3.4.4 IkeWiki ...... 12 4. Research Methods ...... 14 4.1 Project Overview ...... 14 4.2 Objectives ...... 14 4.3 Project Plan...... 16 4.4 Tools and methodologies...... 17 4.4.1 Project stages ...... 17 4.4.2 Tools ...... 17 4.4.3 Implementation...... 18 4.4.4 Testing ...... 18 4.4.5 Extensions...... 19 4.5 Summary...... 20 5. References ...... 21 6. Appendices...... 23 6.1 Appendix A – Wiki Matrix...... 23
2 1. Abstract
This project aims at exploring the use of Semantic Wikis for authoring and reasoning over an OWL DL ontology in a collaborative environment. A number of wiki implementations are here analysed and compared to determine their suitability for extending to this purpose. Semantic wikis allow users to add semantic annotations to wiki content. These annotations provide conceptual or individual information about a wiki page, and its relationships to other pages. The expressivity of annotations varies in most wikis, however this report focuses on OWL ontologies. Using an ontology provides the wiki with an underlying model of its domain’s knowledge, in this case ‘Travel and Holidays’. It is also useful to classify and verify the consistency of the data in the pages, which is done by means of a reasoner. Automated reasoners classify and infer whether new wiki entries are consistent with the knowledge base before storing them.
The implementation proposed in this report consists of enhancing and adapting two semantic wikis to make better use of OWL and reasoning; IkeWiki and BOWiki . The wikis’ functionality will be tested in regards to adding and editing OWL individuals ( ABox ), as well as concepts ( TBox ). Depending on this, either one or both wikis will be adapted to use the Travel ontology, supplied by Dr. Robert Stevens. The ‘Travel wiki’ will subsequently be tested in a collaborative environment, where multiple users author wiki content. This is carried out as outlined in example pages, and also a video demonstration of wiki features. The project will draw conclusions on the usability of the wiki as a collaborative authoring tool for OWL ontologies. The ability to engineer the knowledge contained in an ontology from a semantic wiki brings the added benefits of online collaboration combined with a user friendly web interface. Furthermore, querying and reasoning over an ontology add powerful capabilities that allow a more advanced and widespread use of wikis as support tools for authoring knowledge.
3 2. Introduction
The purpose of this project is to develop a Semantic Wiki, based on a travel ontology provided by Dr. Robert Stevens, which makes extensive use of reasoning both about classes and individuals. The first part of the project, covered in this report, involves a survey of existing semantic media wiki projects and software, in order to make recommendations for the implementation in the subsequent part of the project. This second part of the project entails the installation and testing of a wiki implementation, so as to finally adapt it to make better use of OWL1 (Web Ontology Language) and reasoning.
Over the initial phase of the project several wiki applications have been analysed and compared. The most relevant wiki applications to this project’s goal will be discussed in the Background section, more specifically these wikis should be reusable – so as to extend it to the ontology domain of the project – and provide advanced reasoning capabilities in OWL DL. Furthermore, the intended wiki is meant to be intuitive and user friendly, to the point that collaborative authoring of the knowledge base is as simple and straightforward as possible. If entering content in the wiki is made a complex task, then adherence to a particular style may not be always possible. So in terms of adding and editing entries, the best approach would be wizard or form like interfaces, in order to facilitate what is likely to be the most stressful part of knowledge authoring in wikis.
The first notion of a wiki came to be implemented in 1995 by Ward Cunningham, with the launch of WikiWikiWeb 2. It was then defined as an intelligible piece of software that would provide a simple way to collaboratively describe, author, and collect information [1]. Wikipedia 3 is currently the most popular implementation of a wiki, consisting of a multitude of articles in several languages, and a large online community committed to its development. The notion of community driven contributions is what defines a wiki as a collaborative authoring tool; its pages are written through contributions of not one but several people or organised groups worldwide. Wikipedia is based on the MediaWiki4 wiki engine, which comprises several features to support collaboration; such as recent changes in the wiki, revision history of pages, change tracking, and list of contributions by specific users.
The knowledge contained in a wiki is typically represented in HTML 5 pages, comprised of natural language text and hyperlinks. Having no embedded formal definition of itself or how it relates to other pages and resources, a wiki article has very little machine readable information that one can re use. The querying capability of a regular wiki is an example of this; searching pages in a wiki is done by entering article names (or approximations thereof). It is possible to search Wikipedia for the entry of ‘Athens’ or other city names, but unless such a page has been manually compiled it is not possible to query Wikipedia for a ‘list of all Greek cities’. So despite the amount of information gathered in a regular wiki, it is not structured in a way that allows for article retrieval in this complex way.
1 http://www.w3.org/TR/owl features/ [accessed on 10/02/09] 2 http://c2.com/cgi/wiki [accessed on 19/03/09] 3 http://en.wikipedia.org/wiki/Main_Page [accessed on 05/02/09] 4 http://www.mediawiki.org/wiki/MediaWiki [accessed on 05/02/09] 5 http://www.w3.org/TR/html401/ [access on 17/02/09]
4 On this tone, a Semantic Wiki will be introduced as an extended version of a wiki, with technologies developed by the Semantic Web community, and providing the ability to describe resources in a formal language. The describing of resources, also known as annotating , consists of associating metadata (information about information) with resources in a wiki page. “By adding metadata to ordinary Wiki content, users get added benefits such as improved retrieval, information exchange, and knowledge reuse” [2]. So unlike a regular Wiki, it is possible to author more than just text and links; semantic wikis capture machine readable information about the data in its pages, as well as the relations between them [3]. This way, the knowledge in a semantic wiki can be queried and exported as if it were a normal database, on the grounds that it holds an underlying model of the knowledge contained within its the pages.
Typically, semantic wiki annotations are written in RDF 6, but can also be supplied by an ontology ; a formal model of a shared understanding of a domain, which is interpreted as a set of concepts and relations between these. In the context of the project, an OWL ontology will be used to describe the ‘Travel and Holidays’ domain of the semantic wiki. There are several benefits to using an ontological basis for a semantic wiki; the ability to use automated reasoners, that check if new entries are consistent with the knowledge base before physically storing them, and also powerful querying methods that make use of the reasoner. Below is an example where the content of a wiki page is annotated, using the Semantic MediaWiki 7 Project Halo extension [4]; on the left keywords can be selected, and annotated with further pop up options, or from the right hand side toolbar.
Figure 1 - Annotation mode of Semantic MediaWiki Project Halo extension 8 Adapted from http://www.ontoprise.de/uploads/pics/SMWplus_AnnotationMode.png
6 http://www.w3.org/RDF/ [accessed on 09/03/09] 7 http://semantic mediawiki.org/wiki/Semantic_MediaWiki [accessed on 05/02/09] 8 http://wiki.ontoprise.com/wiki/index.php/Faq/halo_extension [accessed on 19/03/09]
5 3. Background and related literature
3.1 Semantic Wiki approaches
There are numerous semantic wiki implementations publicly available, serving distinguished purposes and with varied functionality. The way in which the content of a wiki is formally represented, annotated, and finally navigated, effectively determines the capabilities of a wiki implementation. Semantic MediaWiki (SMW) [5] is an extension of MediaWiki that provides the ability to annotate the current page. However, it does not detach the page content and the metadata associated with it before storing, thus not taking full advantage of the page annotations for advanced querying and navigation. “The internal knowledge model of Semantic MediaWiki is closely related to OWL DL, although just a small fragment of the expressive means of this language is actually available within the wiki” [6]. Contrary to Semantic MediaWiki, IkeWiki 9 [7, 8] stores the page contents and the annotations separately, and is built upon a background ontology that verifies the consistency and relations of its pages. Similarly, BOWiki 10 employs a core ontology, by default a biological ontology, which provides the background knowledge about types and relations [9]. It is however possible to use an arbitrary core ontology, or access an external ontology from BOWiki’s special pages.
There have been successful attempts at automating the creation of wikis based solely on the input ontology. WikiFactory 11 [10] generates simple semantic wikis from ontological descriptions in OWL; the wiki is automatically populated with pages containing metadata derived from the ontology, as well as the relations among wiki pages. However, in the latest implementation the final output does not provide the complex reasoning and querying methods that other semantic wikis explore, and lacks storage methods for page annotations.
SemperWiki 12 attempts to combine the ease of use virtue of personal wikis with the advanced retrieval and navigation that characterise semantic wikis [11], resulting in a simple and effective personal information management tool. Another example, Powl 13 was initially designed as a web based ontology management tool, which allowed “parsing, storing, querying, manipulating, versioning, serving and serializing RDF and OWL knowledge bases for different target audiences” [12]. Later on Powl was used as the basis for OntoWiki 14 , a wiki like tool that supports agile knowledge engineering in a web based environment. The purpose of OntoWiki is to facilitate the presentation and retrieval of instance data from the knowledge base, by regarding it as an ‘information map’. This allows the user to access ‘map nodes’ using different views; if the selected instance contains geographical data, OntoWiki perceives it as a location on the globe and attempts to match it by calling upon the Google Maps API 15 from within the wiki, and displaying it in a Map View .
9 http://ikewiki.salzburgresearch.at/ [accessed on 11/03/09] 10 http://bowiki.net/wiki/index.php/Main_page [accessed on 11/03/09] 11 http://swe.web.cs.unibo.it/WikiFactory/WebHome [accessed on 17/03/09] 12 http://www.eyaloren.org/semperwiki.html [accessed on 14/03/09] 13 http://sourceforge.net/projects/powl [accessed on 15/03/09] 14 http://ontowiki.net/Projects/OntoWiki [accessed on 15/03/09] 15 http://code.google.com/apis/maps/ [accessed on 16/03/09]
6 3.2 Wiki Architectures
3.2.1 General structure of a Wiki
The typical architecture of a wiki follows the three tier software design model, where the presentation layer (user interface), application layer (process management services), and data layer (data access and storage) are developed and maintained as independent modules [13]. The interaction between these layers is exemplified in Figure 2.
The presentation layer gives the user access to the application; it presents the data to the user and permits data manipulation and entry. This is the top level of the web application, and in the case of Media Wiki – the Wiki engine behind Wikipedia – it communicates with the application layer through HTML requests and dispatching PHP commands.
The application layer contains most of the application logic; the presentation layer presents data to and collects data from the user; the database layer stores and retrieves the data. The application layer executes most of the remaining roles that tie the other layers together; it processes input from the user, forming it into queries on the database to read or write data; drives the structure and content of the data displayed to the user.
The data layer, or database access layer, interacts with persistent data stored in a database. It can be accessed trough the application layer, and in some situations by the presentation layer itself. This is where information is stored and retrieved; the information is then passed on to the application layer for processing, and afterwards redirected to the user via the presentation layer.
Figure 2 – Typical Wiki software architecture Adapted from http://www.mediawiki.org/wiki/Manual:MediaWiki_architecture [Accessed on 05/02/09]
7 3.2.2 General structure of a Semantic Wiki
In comparison with a general wiki, a semantic wiki adds further components and functionality to enable page annotating and storing. When a user edits or creates a new page, the parser analyses the text (typically wiki syntax) and extracts the metadata which is then stored by the ‘page server’. The page server itself acts as a mediator between the other components; namely it calls upon the Analyser to retrieve sets of pages that are related to the current page, displaying it back to the user interface.
In the database layer, it is possible to distinguish between the page content (generally XML content) and the metadata related to it. In this context, the page server is responsible for combining page content with metadata and display as requested.
Figure 3 – Typical Semantic Wiki software architecture [13, 14]
3.3 Types of Semantic Wikis
In the analysis of wikis in Appendix A – Wiki Matrix , RDFS 16 and OWL schemas are often mentioned; in terms of knowledge representation, an OWL ontology is divided in two parts: the assertion box ( ABox ), and the terminology box ( TBox ). “W3C distinguishes between RDF for ABox knowledge and ontology languages like RDFS and OWL for TBox knowledge” [15]. Most of the semantic wikis analysed only offer support for knowledge representation in RDF, but in the interest of this project OWL ABox and TBox authoring are the intended goals. TBox editing in a semantic wiki is associated with functionality of ontology editors, and only a number of wikis offer this, while ABox handling is more widely used. In a general context:
TBox describes classes or concepts of a domain. It is formed by OWL or RDFS 17 , using an extended RDF syntax.
16 http://www.w3.org/TR/rdf schema/ [accessed on 19/03/09] 17 The distinction should be made that OWL is not the same as RDFS; OWL allows all the same expressivity as RDFS and more.
8 ABox describes instances and relationships between instances of concepts. It is typically formed of RDF triples.
From a semantic wiki point of view, authoring a TBox would mean changes at the conceptual level – altering the definition of, for example, Bird from ‘animal that flies’ to ‘animal that flies and has feathers’. This has consequences on the instances of these concepts, if a duck was instantiated as an ‘animal that flies’ then the change implies that when the knowledge base is classified, the duck is no longer regarded as a Bird.
Authoring an ABox in a semantic wiki means changing instances of concepts or the relationships to other instances. From the example above, duck could be altered in order to say that this duck ‘has feathers’, which would re classify duck as an instance of Bird.
3.4 Semantic Wiki implementations
Taking into consideration the semantic wikis mentioned in the Background (and also in Appendix A – Wiki Matrix ), BOWiki, IkeWiki and Semantic MediaWiki have been selected as the most relevant implementation approaches to the project. These three wikis are here compared in terms of usability, functionality, and possible extensions. The table below summarises the main features and information regarding these wikis.
3.4.1 Features Overview
Semantic BOWiki IkeWiki MediaWiki Developers Markus Krötzsch, Joshua Bacher, Michael Sebastian Schaffert, Denny Vrandecic, and Backhaus, Alexandr Salzburg Research others Uciteli, Robert Hoehndorf Purpose Annotating content in Annotating knowledge Ontology engineering RDF (OWL ABox ) via OWL ABox / TBox tool Description A MediaWiki extension A Semantic Wiki for IkeWiki is an easy to that introduces simple collaborative editing of use semantic wiki, annotations to wiki biomedical ontologies with a sophisticated content, and adequate and gene data; it is built user interface and querying and upon an OWL support for importing navigation methods. ontology, with support ontologies and for a DL reasoner reasoning Knowledge OWL DL ontology OWL DL ontology OWL RDFS, OWL representation DL (slower) Implementation PHP PHP JSP Webserver Apache HTTP Server Apache HTTP Server Tomcat Wiki Database MySQL MySQL PostgreSQL Metadata Store MySQL MySQL Jena RDF Store Interaction Wiki syntax, Forms, BOWiki syntax, WYSIWYG, wiki Point and click Simple SemanticSearch syntax License GNU General Public GNU General Public GNU General Public License License License Table 1 - Comparison of wiki implementations
9 3.4.2 Semantic MediaWiki
Semantic MediaWiki (SMW) was initially designed as a simple extension to MediaWiki in order to allow users to annotate the content in its page. The annotations in Semantic MediaWiki are entered into the content of a page using a special mark up, which is then ‘unambiguously’ mapped into the OWL DL knowledge base [16]. According to the documentation, the majority of the annotations are simple ABox statements in OWL syntax, which in this context may describe or classify individuals, or annotate these with data values. However Semantic MediaWiki is unable to handle schematic information in the form of an OWL TBox , since it was not intended as an ontology editor.
In Semantic MediaWiki it is possible to import data from an OWL ABox , as well as exporting formal descriptions of wiki pages in OWL/RDF format; which means the whole ontology can be exported in the same way. The ability to import and export parts of an ontology allows for reuse of the information in the wiki pages or the overall domain represented. The content in Semantic MediaWiki can be queried using the same syntax as for annotating; queries are done via a special page Special:Ask , or via inline queries. Finally it is possible to save conceptual queries in a wiki page, which act as dynamic queries in the page content.
Figure 4 - Semantic MediaWiki Ontology browser, Adapted from http://sourceforge.net/dbimage.php?id=162688 [accessed on 08/04/09]
On top of these aspects, Semantic MediaWiki has a number of issues that would constrain the implementation of this project. In particular, it does not support a number of ontology features: transitivity, inverse properties, domain and range restrictions, number restrictions and functional properties. This poses a major restriction in terms of ontological expressivity. Reasoning capabilities for OWL DL in Semantic MediaWiki have to be configured separately, but would be limited nonetheless. This is because the wiki stores both the page content and its annotations together, with no formal distinction. Such distinction would make reasoning over metadata possible, as well as increasing the overall performance of page retrieval and querying; since the entire page content and metadata have to be parsed and the relevant code retrieved for each request.
10 3.4.3 BOWiki
The BOWiki is a collaborative editor that extends MediaWiki by using an ontology to validate all the semantic information related to the wiki pages. Originally it was intended as a biomedical wiki, particularly dedicated to gene functions. But in spite of this, BOWiki allows importing an arbitrary OWL ontology to replace the original core ontology. The wiki is comprised of four main components [17], which are also depicted in Figure 7:
The BOWiki software extension – the main application component, a presentation layer addition to MediaWiki.
BOWiki database extension – an enhancement of the MediaWiki data layer for separate semantic data storage.
BOWikiserver, based on the Jena 2 Semantic Web Framework [18] and the Pellet reasoner [19] – classifies and reasons over the content of the wiki’s knowledge base.
OWL DL ontology.
Figure 5 - Introducing a new relation in BOWiki [20] Figure 6 - Inconsistency detection entering data
The BOWiki extension to the MediaWiki is responsible for processing the semantic annotations entered in the wiki pages, which are then transferred to the BOWikiServer. At this point the semantic data is checked for its consistency with the resident OWL ontology (which has to be pre loaded onto the wiki at installation time). This check is carried out by the description logic reasoner, via the Jena 2 Semantic Web framework, which will then commit the semantic data to storage if it is consistent.
The page content and its related annotations are stored separately; the page content is stored in the MediaWiki database, while the annotations are stored in the BOWiki database extension. This annotation database extension uses separate tables for storing subjects, relations, roles, types and objects [21]. The database model of the wiki is divided into two semantic store parts; the first is dedicated to classes relations, roles and types, and the second to instances instances of relations, roles or objects.
11
Figure 7 - BOWiki software architecture adapted from http://onto.eva.mpg.de/pub/eswc-misc/ [accessed on 06/04/09]
3.4.4 IkeWiki
The development of IkeWiki started as a support tool for knowledge workers to collaboratively formalise knowledge. Currently the wiki supports importing and exporting of a knowledge base in RDFS and OWL. The basic user interface and functionality of IkiWiki resembles MediaWiki, but it is an AJAX application based on the Dojo Toolkit 18 . Editing content is facilitated by a WYSIWYG editor, which includes AJAX based Link and Page Annotations. The architecture of IkeWiki, in Figure 8, includes three main components; the Rendering Pipeline, RDF Store and Page Store.
Figure 8 - IkeWiki software architecture [7]
18 http://dojotoolkit.org/
12 Page Store; stores and retrieves the content of wiki pages from the database. The page content is represented in Wiki Interchange Format (WIF).
RDF Store; based on the Jena 2 Framework for representation of the knowledge base, includes a SPARQL engine that enables querying of the KB.
Rendering Pipeline; responsible for combining the page content, in the XML based WIF format, with the semantic data associated with it.
Despite the fact that IkeWiki supports OWL, the reasoning support is by default limited to RDFS. In order to add an OWL DL reasoner such as Pellet some changes need to be done to the setup and code, which is part of the research to be ensued. Enabling OWL DL reasoning, however, could reportedly slow down the wiki performance to a certain degree, which may be related to the processing and storage of the ontology.
Figure 9 - IkeWiki example page content and functionality [7]
13 4. Research Methods
4.1 Project Overview
In the Background section several wiki systems were described in relevant detail, and from their functionality there are two appropriate candidates to extend the initial specification of the project:
BOWiki due to its complex reasoning and querying capabilities IkeWiki for its advanced user interface, and similar functionality to BOWiki.
From the documentation available both wikis allow importing OWL ontologies as the core ontology for the wiki pages, which is the main concern for extending an existing implementation. Moreover the reasoning and querying tools used in these wikis are adequately complex for the goals of this project; therefore it seems sensible to pursue installing and adapting both systems concurrently, to evaluate their suitability in relation to the ‘travel and holiday’ domain.
In the initial implementation and testing stages of the project both wikis will be compared simultaneously, and it is possible that only one will proceed to the following implementation and testing stages. Since reasoning in IkeWiki is limited to RDFS by default, the issue that arises is adapting the wiki to use the Pellet reasoner. In case the tests carried out when reasoning over OWL DL in IkeWiki are successful, then it would be most suitable for the project’s purposes. Otherwise the remaining three objectives of the project would apply only to BOWiki.
4.2 Objectives
Based on the analysis of previous approaches to semantic wikis and the purpose set forth in the Introduction , the main objectives of the project are compiled in the table below, along with the tasks involved. The assignment of tasks in relation to each major objective is represented on the Gantt chart in the following section.
Main objectives Tasks Examine wikis functionality Configure wiki to import OWL ontologies with an OWL ontology Employ an OWL DL ontology as main knowledge base Adapt reasoner to reason over the ontology Examine the generated articles in the wiki pages
Test usability and Verify accurate classification of the ontology performance of both wikis Identify if wiki pages are populated accordingly Evaluate which implementation provides a more user friendly interface Compare the wikis’ handling of ontology changes: ABox and TBox
14 Implementation of the Determine if changes need be made to the ontology Travel ontology wiki Import the travel ontology onto the wiki Verify accurate classification of the ontology Identify if wiki pages are populated accordingly Check that adding and editing wiki entries commits the appropriate ontological changes
Test wiki implementation in Produce a script for entering data in wiki pages collaborative environment Produce a video demonstration of how to add and annotate content Arrange for a number of users to submit any number of entries, according to the script and video Ensure the validity of information entered by the users in the wiki Verify the produced script is being adhered to
Explore extensions to the Analyse user interface extensions to facilitate implementation (if authoring; Forms, WYSIWYG remaining project time Determine the usability of Google Maps within the allows for this) wiki (Semantic Google Maps extension)
Final testing of the wiki Ensure correct functionality of all components if new extensions are added Verify accurate progress of the wiki ‘population’
Table 2 - Main goals and tasks involved in the project
15 4.3 Project Plan
16 4.4 Tools and methodologies
Following the objectives described before, the project will be implemented and tested over five stages; at each major milestone seen in the Gantt chart there is a testing period, to ensure correct functionality and accordance to the requirements before proceeding to the next implementation or test stage. Aside from the final testing, the major test phase is the collaborative editing of the wiki. At this point, several users will be engaged in entering content in wiki pages following a pre established approach, which will be discussed in Testing Stages . The division of project stages derived from the project objectives and the Gantt chart is summarised below.
4.4.1 Project stages
Stage Duration 1.1 Implementing wikis with imported ontology 2 weeks 1.2 Testing wiki implementations 2 weeks 2. Implementing wiki with Travel ontology 1 week 3. Collaborative testing of the wiki 3 weeks 4. Implementing extensions 1 week 5. Final testing 1 week 6. Dissertation writing 4 weeks Table 3 - Stages of the project
4.4.2 Tools
The software tools required throughout the project include support for ontology editing, programming in Java and in PHP. Table 4 illustrates the software that will be used.
Software Developers Usage Protégé 4.0 19 CO ODE OWL ontology editing Eclipse IDE for Java EE Eclipse Foundation Java/JSP development Eclipse Foundation, PHP development tools PDT 2.0 Eclipse plug in 20 Zend Technologies framework Apache Foundation, LAMP (Linux Apache HTTP web server, database The PHP Group, Sun MySQL PHP) server management system, and PHP Microsystems XVidCap 21 Karl Beckers Screen casting software Apache Software HTTP web server for Java Apache Tomcat Foundation Servlets and JSP Pellet reasoner 22 Clark & Parsia OWL DL reasoning Table 4 - Software tools involved in the project
19 http://protege.stanford.edu/ 20 http://www.eclipse.org/pdt/ 21 http://xvidcap.sourceforge.net/ 22 http://clarkparsia.com/pellet
17 4.4.3 Implementation
Phase 1: Implementing wikis with imported ontology
At this stage, the intention is to compare the usability and performance of IkeWiki and BOWiki with an arbitrary OWL ontology. The wikis need to be installed and configured so that a test ontology can be imported and evaluated. From the documentation and user manuals available BOWiki provides this functionality by default. IkeWiki however, poses a bigger obstacle as its reasoning capabilities are not suitable for OWL DL. So the Pellet reasoner will be adapted to IkeWiki – which according to the developer is possible 23 and supported by the wiki. At the end of this phase it is expected that both wikis will be making full use of reasoning over the test ontology, so that conclusions can be drawn in regards to performance during the testing stage (stage 1.2 of Table 3).
Phase 2: Implementing wiki with Travel ontology
Based on the tests carried out with the initial ontology, the implementation of the Travel wiki will proceed on either one or both wikis. The system that shows better handling of OWL ABox and TBox is preferable for continuing the project.
The Travel and Holiday ontology will then be imported onto the wiki, and upon classifying the new population will be generated from the knowledge base. At this point, the performance of the wiki will be reviewed, with regards to reasoning and interfacing within the wiki.
4.4.4 Testing
Phase 1: Testing wiki implementations