Pay-as-you-go Ontology Query Answering Using a Datalog Reasoner

We describe a hybrid approach to conjunctive query answering over OWL 2 ontologies that combines a datalog reasoner with a fully-fledged OWL 2 reasoner in order to provide scalable “pay as you go” performance. Our approach delegates the bulk of the computation to the highly scalable datalog engine and resorts to expensive OWL 2 reasoning only as necessary to fully answer the query. We have implemented a prototype system that uses RDFox as a datalog reasoner, and HermiT as an OWL 2 reasoner. Our evaluation over both benchmark and realistic ontologies and datasets suggests the feasibility of our approach.