Personalized Exploitation of XML Documents Based on a Domain Ontology

Abstract In this paper, we propose an approach for personalized information retrieval of semi-structured documents (XML). The central idea of our work is that the use of external semantic resources such as ontologies (Domain Ontology) and user profile can improve the efficiency of the search process. The weighting of the concepts of our developed domain ontology allows to give more importance to the most specific concepts, because the lower the level at which the concept is located in the hierarchy, the more detailed and targeted the information it provides. Also, the use of a user profile is decisive for the result of a search, without it, an Information Retrieval System (IRS) will behave exactly the same way with all users, i.e. the IRS will return the same list of results for the same query submitted by users with different information needs: they have different knowledge, different preferences, as well as different goals and interests. In the literature, several user profile models have been proposed. In our approach, we represent the user profile as a vector of weighted concepts, as well as documents and queries. We also propose to use semantic similarity measures for relevance assessment. The results obtained using this approach are very promising, which demonstrates the contribution of using ontologies and user profile in improving the efficiency of the search process.