ABSTRACT
Ontologies form the basis of the semantic web. They represent a formal specification of concepts and relationships in a domain of interest. As the semantic web develops, there is a growing number of ontologies publicly available online. This fast growth of ontologies creates difficulties in using, understanding, and managing them. In addition, problems (also called anomalies) in the representation of ontology structures, such as circularities, redundancies, inconsistencies, and deficiencies, may occur during the construction or integration of these ontologies. There is a need for techniques that help users to evaluate ontology quality and uncover such anomalies. To tackle this problem, we present an ontology to describe ontology structures, called MetaFOR, and a system that uses it to spot anomaly occurrences, called ONTO-Analyst. The system generates a structural description of an ontology, using MetaFOR, and then queries this description, using SPARQL queries, to find anomalies. The ONTO-Analyst system was tested using a representative set of 18 anomaly types, taken from the literature, in a set of 608 OWL ontologies from four major public repositories. The tool found more than 3 million occurrences of 12 anomaly types: three circularities, five redundancies, one inconsistency, and three deficiencies. The testing results demonstrate that the ONTO-Analyst method is capable of automatically identifying many 292kinds of anomalies and that even widely used ontologies have some anomaly issues.
