A multi‐objective particle swarm optimization for matching domain ontologies

Abstract

As a generic conceptual model for describing domain knowledge, ontology is able to provide a formal and concrete description on the concepts in different domains. Since different ontology engineers have diverse viewpoints on the world, leading to the heterogeneity issue among different ontologies. To solve this problem, this paper investigates how to make certain the correspondences of entities between different domain ontologies, that is, domain ontology matching problem. When solving this problem, how to effectively aggregate different similarity measures to differentiate entities and obtain high‐quality alignment is the kernel issue. In this paper, we first introduce two statistical metrics for measuring ontology alignment's quality, then, an optimization model is established for the multi‐objective ontology matching problem, and a Multi‐Objective Particle Swarm Optimization algorithm (MOPSO) is proposed to optimize the quality of domain ontology alignment. The experimental results illustrate that it outperforms OAEI's participants with respect to both recall and precision.