Towards richer relation semantics on the Semantic Web.

摘要

Knowledge representation formalisms for the Semantic Web have inadequate capabilities with respect to specifying relation semantics. This dissertation presents implemented techniques to add relation semantics to Semantic Web ontologies. First, syntactical constructs have been developed to specify inter-link constraints on sets of relations in ontologies. The semantics of these constructs specifying mutual exclusiveness and collective exhaustiveness upon relation instances can be enforced using algorithms implemented in Prolog that are presented in this dissertation. To capture the intrinsic semantics of relations from the perspective of the Semantic Web, a two-tier framework of relation elements (or simply elements) has been developed. Constructs to specify these elements and their values, which describe the intrinsic nature of relations in ontologies, have also been developed. The OWL Enhance prototype has been developed in this dissertation to use these constructs to add richer relation semantics to input ontologies implemented in the Web Ontology Language (OWL) and demonstrate their usefulness in creating relation taxonomies and clusters. The OWL Enhance prototype has three modules: the Relation Semantics Elicitation Prototype (RSEP), the Relation Clustering Module (RCM) and the Relation Taxonomy Module (RTM). RSEP interactively elicits additional intrinsic relation semantics in the form of descriptive element values from knowledge providers and appends them to input OWL ontologies. RTM and RCM use these elicited relation semantics to taxonomically arrange and cluster relations respectively, which may contribute to the cause of knowledge sharing and reuse on the Semantic Web. Results from testing the OWL Enhance prototype on domain specific and high-level ontologies are presented in this dissertation. These results are discussed in the context of the potential utility of richer relation semantics to the cause of greater accuracy of knowledge representation in ontologies and the reuse and interoperability of knowledge on the Semantic Web.