Section: Software

The OntoSim library

Participants : Jérôme David [Contact] , Jérôme Euzenat.

OntoSim is a library offering similarity and distance measures between ontology entities as well as between ontology themselves. It materialises our work towards better ontology proximity measures.

There are many reasons for measuring a distance between ontologies. For example, in semantic social networks, when a peer looks for a particular information, it could be more appropriate to send queries to peers having closer ontologies because it will be easier to translate them and it is more likely that such a peer have the information of interest. OntoSim provides a framework for designing various kinds of similarities. In particular, we differentiate similarities in the ontology space from those in the alignment space. The latter ones make use of available alignments in an ontology network while the former only rely on ontology data. OntoSim is provided with 4 entity measures which can be combined using various aggregation schemes (average linkage, Hausdorff, maximum weight coupling, etc.), 2 kinds of vector space measures (boolean and TF.IDF ), and 4 alignment space measures. In addition, the framework embeds external similarity libraries which can be combined to our owns.

This year, we have implemented the measures of agreement and disagreement between ontologies recently proposed by Mathieu d'Aquin (Open university).

OntoSim is based on an ontology interface allowing for using ontology parsed with different API s.

OntoSim is written in Java and is available under the LGPL license at http://ontosim.gforge.inria.fr .