An Improving Mapping Process Based on a Clustering Algorithm for Modeling Hybrid and Dynamic Ontological User Profile

The main element of a personalization's process of information retrieval is the user. However, the effective modeling of the user profile remains a major challenge. In this paper, we propose an algorithm for mapping the web pages visited by a user during his interactions on the web with a reference ontology. Based on this mapping algorithm, we propose a dynamic and hybrid approach of the user profile, able to detect user's interests and the changes in these interests. By comparing our mapping algorithm to three other mapping approaches, we show that our algorithm has a better performance than these different algorithms. Moreover, the experimentation of our proposed multi-agent system confirms that our model can detect dynamically the user's interests and the changes in these interests.