An efficient model for extracting an optimal alignment with multiple cardinalities in ontology alignment
Chahira Touati, Moussa Benaissa, Yahia Lebbah · International Journal of Metadata Semantics and Ontologies · 2016
Ontologies have been created to solve the data heterogeneity problem on the web and to share domain knowledge between systems. However, these ontologies have become themselves a source of heterogeneity. The ontology alignment is an effective solution to this problem, which is to discover the semantic correspondences between the ontologies entities, so it becomes a crucial task. This paper proposes an efficient flow-based algorithm which maximises the global similarity objective function subject to multiple cardinality constraints. Our approach has been evaluated on a variety of synthetic and real data, and compared with currently used algorithms (e.g. Hungarian and Karp algorithms). The obtained results show the efficiency of our approach, especially in the case of rectangular matrices, which is the most frequent case in practice (real ontologies).