Implementing an InferenceEngine forRDFS/OWL Constructs and User-DefinedRulesinOracle

ThisInference Enginesarean integral partof Semantic DataStores. Inthis paper, wedescribe ourexperience ofimplementing ascalable inference engine forOracle Semantic DataStore. Thisinference engine computes production rule basedentailment ofoneormoreRDFS/OWLencoded semantic datamodels.The inference enginecapabilities include i) inferencing basedonsemantics ofRDFS/OWLconstructs and user-defined rules, ii) computing ancillary information (namely, semantic distance andproof) forinferred triples, andiii) validation ofsemantic datamodelbasedon RDFS/OWL semantics. A unique aspect ofourapproach isthattheinference engine isimplemented entirely asadatabase application ontop ofOracleDatabase. The paperdescribes theinferencing requirements, challenges insupporting asufficiently expressive setofRDFS/OWLconstructs, andtechniques adopted tobuild a scalable inference engine. A performance study conducted using bothnative andsynthesized semantic datasets demonstrates the effectiveness ofourapproach.