Ontology Alignment using Stable Matching

Abstract

Ontology matching is a solution to the problem of semantic heterogeneity in the integration and sharing of information. It consists in establishing mappings between entities which semantically belong to different ontologies. Most ontology mapping approaches use elementary matching techniques (eg, string-based methods, linguistic methods, etc.). These techniques map the elements by analyzing the entities in isolation and ignoring their relationships with other entities (father/son, brother, etc.). Bypassing the latter aspects, the determination of the semantics of an entity is often difficult. Hence, the structural information of an ontology plays an important role in ontologies mapping. In this work, we study methods of structural alignment. We adopt two alignment methods that are based on the structure. The first, called Method of Similarity of Inheritance (MSI), applies the initial similarity method based on the concepts and integrates the inheritance relation (father/son) into the calculation of similarity. The second, called Method of Sibling Similarity (MSS), involves sibling relationships to enrich the similarity score between two concepts. Moreover, we enhance the mappings selection strategy by integrating the stable marriage algorithm as an optimal matching strategy. This algorithm improves the ontology alignment quality through efficient optimization techniques and creates a faster and more robust alignment method.