Various data sources on the Web tend to be highly dynamic; this is evident in prominent Web services frameworks in which devices register or deregister their descriptions quite rapidly and in Semantic Web portals which allow content authors to modify or extend underlying ontologies and submit content. Such applications often leverage Description Logic (DL) reasoning for a variety of tasks (e.g., classifying Web service descriptions, etc); however, this can introduce substantial overhead due to content fluctuation, as DL reasoners have only been considered for relatively static knowledge bases. This work aims to provide more efficient DL reasoning techniques for frequently changing instance bases (ABoxes). More specifically, we investigate the process of incrementally updating tableau completion graphs used for reasoning in the expressive DLs SHOQ and SHIQ, which correspond to a large subset of the W3C standard Web Ontology Language, OWL-DL. We present an algorithm for updating completion graphs under the syntactic addition and removal of ABox assertions. We also provide an empirical analysis of the approach through an implementation in the OWL-DL reasoner, Pellet.
展开▼
