Multiobjective particle swarm optimization based ontology alignment

Abstract

Ontology alignment plays a vital role for interoperability among the heterogeneous semantic data sources. It is a set of correspondences between two or more ontologies. There are lots of methods to measure the semantic similarity between entities from several ontologies. To acquire the comprehensive and precise results, all the similarity measures are integrated. Therefore, integrating different similarity measures into a single similarity metric pose a challenging problem. In general, weights corresponding to various similarity measures are assigned manually or through some method. The problem is that it suffers from lack of optimality. There are many evolutionary based approaches to find the optimal solution but they optimize a single objective function. In this article, a multiobjective particle swarm optimization algorithm is proposed for achieving various weights correspond to different similarity measures. Then subsequently similarity aggregation function is calculated for identifying the optimal alignment. Here, two objectives precision and recall are simultaneously optimized and a optimal alignment is produced for which f-measure is very high.