Fuzzy Ontology-Based Approach for Automatic Construction of User Profiles

This paper shows a fuzzy ontology based approach to automatically build user profiles from a collection of user interest documents. The ontological representation of the user profile enhances the performance in tasks such as filtering, categorization and information retrieval. The proposed technique takes advantage of relevance measures to generate semantic representations of user context. The proposed work also presents a strategy for automatic generation of fuzzy ontologies to support user profile modeling. The experiments performed confirm that the automatically obtained fuzzy ontologies are good representation of the user’s preferences. In order to test the applicability of the obtained ontologies, a text categorization experiment has been proposed and the obtained results indicate that the approach can be applied with satisfactory results and warrants further research.