Advancing the Enterprise-class OWL Inference Engine in Oracle Database

OWL is a standard ontology language defined by W3C that is used for knowledge representation, discovery, and integration. Having a solid OWL reasoning engine inside a relational database system like Oracle is significant because 1) many relational techniques, including query optimization, compression, partitioning, and parallel execution, can be inherited and applied; and 2) relational databases are still the primary place holder for enterprise information and there is an increasing use of OWL for representing such information. Our approach is to perform data intensive reasoning as close as possible to the data. Since 2006, we have been developing an RDBMS-based large scale and efficient forward-chaining inference engine capable of handling RDF(S), OWL 2 RL/RDF, SKOS, and user defined rules. In this paper, we discuss our recent implementation and optimization techniques for query-rewrite based OWL 2 QL reasoning, named graph-based inference (local inference), and integration with external OWL reasoners.