Approximating OWL-DL Ontologies

Efficient query answering over ontologies is one of the most useful and important services to support Semantic Web applications. Approximation has been identified as a potential way to reduce the complexity of query answering over OWL DL ontologies. Existing approaches are mainly based on syntactic approximation of ontological axioms and queries. In this paper, we propose to recast the idea of knowledge compilation into approximating OWL DL ontologies with DL-Lite ontologies, against which query answering has only polynomial data complexity. We identify a useful category of queries for which our approach guarantees also completeness. Furthermore, this paper reports on the implementation of our approach in the ONTOSEARCH2 system and preliminary, but encouraging, benchmark results which compare ONTOSEARCH2's response times on a number of queries with those of existing ontology reasoning systems.