Inconsistent ontology revision based on ontology constructs

Abstract

Assessing the quality of ontologies is important to reuse/integrate ontologies. In this paper, we propose an approach to revise ontology inconsistency by measuring the relevant ontology constructs such as semantic partitions, axiom fanouts per concept, minimal incoherence-preserving subsets, incoherence impact value per axiom and preferable revision weight per concept in the ontology being measured. We also propose several algorithms to compute the measurement values of these ontology constructs, which can be used to pinpoint and rank which axioms and concepts are most likely to cause the ontology inconsistency. Our approach can help determine which concepts should be preferentially modified and therefore enables automatic revision of ontology inconsistency in the context of a dynamic and evolving environment.