Explaining Inconsistencies in OWL Ontologies

Justifications play a central role as the basis for explaining entailments in OWL ontologies. While techniques for computing justifications for entailments in consistent ontologies are theoretically and practically well-understood, little is known about the practicalities of computing justifications for inconsistent ontologies. This is despite the fact that justifications are important for repairing inconsistent ontologies, and can be used as a basis for paraconsistent reasoning. This paper presents algorithms, optimisations, and experiments in this area. Surprisingly, it turns out that justifications for inconsistent ontologies are more “difficult” to compute and are often more “numerous” than justifications for entailments in consistent ontologies: whereas it is always possible to compute some justifications, it is often not possible to compute all justifications for real world inconsistent ontologies.