Isolation-based subsumption reasoning with enormous volume of web ontologies for scalable semantic service discovery

摘要

The description of service with expressive Web Ontology Language (OWL) ontologies facilitates semantics-based service advertisement and discovery. However, modern description logic (DL) reasoners do not provide sufficiently tolerable and scalable infrastructure for the advertisement and discovery of service profiles on the basis of enormous volumes of OWL ontologies on the web. In this paper, we present an import graph-based tolerable and scalable reasoner that can be implemented on top of off-the-shelf DL reasoners. The reasoner has been designed to reduce the reasoning workload of these reasoners by excluding ontologies not relevant for resolving a given subsumption query. On average, the proposed method isolates about 87.6%-99.9% of all ontologies when performing a given reasoning task. This leads to a significant reduction, by at least 85.6%, in the response time of subsumption reasoning when a naive reasoning strategy involving the loading of all ontologies onto a reasoner is used.