IJCAI-09 Workshop in Identity and Reference in web-based Knowledge Representation

The OKKAM-sponsored workshop on Identity and Reference in web-based Knowledge Representation (IR-KR2009) has been accepted at the 21st International Joint Conference on Artificial Intelligence (IJCAI-09), which will be held in Pasadena on July 11-13, 2009.

Below is a preliminary description of the workshop topics:

Identity and Reference in web-based Knowledge Representation (IR-KR 2009)

The Semantic Web initiative has brought forward the idea that the web may become a space not only for publishing and interlinking documents (through HTML hyperlinks), but also for publishing and interlinking knowledge bases (e.g. in the form of RDF graphs) in an open and fully decentralized environment. This is how Tim Berners-Lee expressed this idea in a note from 1998:

"The Semantic Web is what we will get if we perform the same globalization process to Knowledge Representation that the Web initially did to Hypertext" [see http://www.w3.org/DesignIssues/RDFnot.html]

Even though models and languages to achieve this goal have been taken from long-standing research in AI, it is important to remark that the priorities are different. While traditionally the focus has been on theories to support sound and complete reasoning, web-oriented KR is primarily aiming at dealing with issues of web-wide informationinteroperability and integration. With respect to this, perhaps the most central issues is Principle of Global Identifiers: "global naming leads to global network effects" (see Architecture of the World Wide Web, Volume One, 2004, at

http://www.w3.org/TR/2004/REC-webarch-20041215/). In other words, if a resource (where a resource may range from concrete to abstract objects, from particulars to universals) is globally identified through a uniform identifier in any knowledge repository exposed on the web (e.g. in an RDF store), then any knowledge about it would be much easier to gather and integrate, distributed reasoning becomes practically possible, and knowledge-based navigation acrossinterlinked knowledge sources can be enabled. As it happened for the web of documents, the overall value of such an open and distributed network of interlinked knowledge sources would be immensely bigger than the sum of the value of the components.

Technically, URIs (Uniform Resource Identifiers, see http://www.w3.org/Addressing/) are used to identify entities on the

Semantic Web, but how to achieve shared URI understanding and reuse is object of research. This central role of identity and reference for a web-scale KR poses new challenges to traditional KR, and many researchers have suggested that the concept of URI may deeply affect the notions of language (e.g. the semantics of using the "same" URI in different models), reference (e.g. rigid vs. non rigid designation), interpretation (e.g. the meaning of "links" across knowlkedge bases) & reasoning (e.g. distributed reasoning across theories) in traditional logic-based KR in AI. The goal of this workshop, which in its past editions was mainly restricted to the Web and Semantic Web communities (see past editions at WWW2006, WWW2007, ESWC2008), is to open the debate on the impact and the challenges that web-oriented KR poses to some of the core concepts of traditional AI. Therefore, the workshop aims at collecting contributions which can roughly be grouped as follows:

• Foundations: formal and conceptual theories of identity and reference for web-oriented KR

• Formal theories: semantics for KR on the web, soundness and completeness of web-oriented reasoning, semantics of interlinked data

• Vision papers: visionary solutions to the problems of identity and reference in KR

• Project papers: descriptions of research & development projects in this area

• Experiences: contributions from research and industry that illustrate  case studies or approaches to deal with the issues of identity and reference on a web-scale

• Critical viewpoints: discussions of advantages and disadvantages of the proposed approaches

We especially encourage contributions from groups or organizations which are working on putting together big knowledge-based data collections in order to compare the different practical solutions which were found for integrating semantic data from different sources.