The Open World: Reference and Meaning on the Web
Harry Halpin · 2008
This thesis examines the question: What does a URI mean? Or in other words, what does a URI identify and refer to? This question is of utmost importance for the creation and re-use of URIs on the Semantic Web. An philosophical and historical introduction to the Web explains the primary purpose of the Web as a universal information space for naming and accessing information. A terminology, based in widely-known philo-sophical distinctions, is employed to define precisely what is meant by information, language, representation, and digitality. These terms are then employed to create a foundational ontology and principles of Web architecture, which can be used to de-limit the boundaries of the Web. From this perspective, the Semantic Web is then viewed as the application of the principles of Web architecture to knowledge repre-sentation. However, the classical philosophical problems of identity, reference, and meaning that traditionally caused problems for knowledge representation are inherited by the Semantic Web. Three main positions are inspected: the logicist position, as exemplified by formal semantics and a neo-Russellian descriptivist theory of names, the direct reference position, as exemplified by Putnam and Kripke’s causal theory of reference, and a neo-Wittgensteinian position that views the Semantic Web as yet an-other public language. After identifying the neo-Wittgensteinian viewpoint as the most promising, a solution of using people’s everyday use of search engines is proposed as a proper neo-Wittgensteinian way to determine the meaning of URIs via sense identi-fication and disambiguation. This solution is then tested on a corpus of search terms from a real-world major search engine query log, and these terms are used to retrieve a corpus of Semantic Web URIs. Human browsing of hypertext web-pages is then used as a feature set in order to disambiguate and identify the “sense ” of these URIs via a variety of machine-learning techniques. The results show that @@. Future work for the Semantic Web that follows from our argument is detailed, both of a technical and conceptual nature. iii Acknowledgements