Knee Solution-Driven, Decomposition-Dased Multi-Objective Particle Swarm Optimization for Ontology Meta-Matching

As the latest information exchange model, ontology is favored by information systems, but the heterogeneity of ontology has seriously influenced the interaction and cooperation between these systems. Ontology matching is considered an effective method to solve the ontology heterogeneity problem whose kernel technology is a similarity measure. However, a single measure cannot achieve satisfactory ontology alignments. To this end, integrating different similarity measures is feasible. First of all, due to the difference in user preferences for alignment quality, the ontology matching problem is modeled as a continuous multi-objective optimization model. Particle Swarm Optimization (PSO) is suitable for solving continuous optimization problems and previous studies have found that decomposition-based methods are more suitable for solving ontology matching. Then, considering the user’s preference, a knee solution-driven, decomposition-based multi-objective particle swarm algorithm (K-MOPSO/D) is designed to solve the ontology matching. Finally, the effectiveness of our proposed method is verified by standard test cases from the well-known OAEI (Ontology Alignment Evaluation Initiative), and its performance is compared with the state-of-the-art matching methods.