A Behavioral Metrics Suite for Modular Ontologies

Abstract

Modular ontologies are superior to monolithic ontologies as they promote reusability. Current work proposes a behavior metrics' suite to measure the quality of modular ontology and hence its reusability. It is imperative to assess a modular ontology, not merely from the point of view of its structure but also on the ground of its intrinsic characteristics that bear on its behavior, significantly. Relevant literature has been analyzed for ontology metrics that define the knowledge embedded in ontology. Knowledge encapsulation, relationship and concept coverage, and depth of subsumption hierarchy have been keyed out as important measures for assessing the modular ontology's behavior, which in turn affects the Semantic Web application it is affiliated with. The suggested measures are unambiguous and can be put on to whatever type of ontology, such as, cross-domain, domain, application or task ontology, in order to facilitate the assessment of quality of ontology or a partition thereof. The IEEE 1061 framework has been used to validate the usefulness and correctness of the proposed metrics. A prototype that implements our metrics has been proven, it computes the overall behavioral quality of ontology by using the weighted scale method on the component behavioral metrics.