An Experiment with Ontology Mapping using Concept Similarity

This paper describes a system for automatically mapping between concepts in different ontologies. The motivation for the research stems from the Diogene project, in which the project's own ontology covering the ICT domain is mapped to external ontologies, in order that their associated content can automatically be included in the Diogene system. An approach involving measuring the similarity of concepts is introduced, in which standard Information Retrieval indexing techniques are applied to concept descriptions. A matrix representing the similarity of concepts in two ontologies is generated, and a mapping is performed based on two parameters: the domain coverage of the ontologies, and their levels of granularity. Finally, some initial experimentation is presented which suggests that our approach meets the project's unique set of requirements.