A plausible inference prototype for the Semantic Web

摘要

This paper presents a prototype that is capable of drawing plausible inferences from Resource Description Framework (RDF) assertions that are constituents of a distributed, Semantic Web knowledge system. The approach taken to build the prototype can be viewed as the extension and adaptation of a classical approach to plausible inference to exploit the evolving infrastructure being developed to represent declarative knowledge on the Semantic Web. The approach includes a knowledge representation formalism that supports meta properties, which define precise semantics, enabling subsequent plausible inferences via extended composition of RDF properties. Most research and development in the context of the Semantic Web has been devoted to representational infrastructure and accompanying query and logical deduction formalisms to evolve the Web from a document repository to a set of distributed knowledge bases. The work presented in this paper provides a functioning Semantic Web application in which the generation of new inferences is not contained within the deductive closure of the knowledge and data expressed by a collection of information sources represented using RDF. Moreover, the paper provides a concrete example of an RDF schema and a working system built around it which demonstrates one potential use of meta data on the Web.