Ontology mapping using bipartite graph


SEÇER A., Coskun Sonmez A., Aydin H.

International Journal of Physical Sciences, vol.6, no.17, pp.4224-4244, 2011 (Peer-Reviewed Journal) identifier

  • Publication Type: Article / Article
  • Volume: 6 Issue: 17
  • Publication Date: 2011
  • Journal Name: International Journal of Physical Sciences
  • Journal Indexes: Scopus
  • Page Numbers: pp.4224-4244
  • Keywords: Bipartite graph, Kuhn-Munkres optimal assignment algorithm, Levensthein metric, Ontology, Ontology mapping

Abstract

Fast improvement of web technologies have caused a problem of semantic integration between distributed applications. In this respect, sharing and distribution of information are of vital importance for ontologies. Ontologies may be improved by any independent association and still be used by another association. In case an association decides to use an ontology improved by another association, they should make mapping between ontology concepts. Different worldviews assign different meanings to different concepts and defines them differently. Therefore, mapping stands as an inevitable process. Mapping is the job of finding objects that are compatible between two ontologies. Semantic mapping of ontologies by means of bipartite graph matching algorithms has been studied in this paper. A mapping system defined as arg Osource →MOtarget has been improved. We have named this system BGOM (bipartite graph optimal mapping). BGOM system finds the one-to-one matching between ontologies Os and Ot which are in similar domains or in same domain. Firstly, two data matrices for ontology concepts Os and Ot have been obtained. Next, a score matrix has been obtained from general data matrixes by using Levensthein metric. Finally, Kuhn-Munkres optimal assignment algorithm has been used to optimally map the concepts between Os and Ot. The reason for this is to find one-to-one matches of concepts in the model we have improved. Kuhn-Munkres algorithm is an effective way to find the most similar couples (Tassa, 2007). Consequently, a one-to-one optimal map has been obtained between source and target ontologies. Application, prediction capability and truth values of BGOM system is evaluated by ontology alignment evaluation initiative (OAEI1) and satisfactory results have been obtained. Precision, recall and f-measure values of alignment results in the system we have improved, and are compared to the other systems in OAEI campaign and considerably good results have been obtained. Application of BGOM system between source and target ontologies has assisted effectively for solution of ontology mapping problem. © 2011 Academic Journals.