Résumé

Pour satisfaire les besoins d’informations d’un utilisateur et la personnalisation de l’information,

Abstract

To meet user needs of information and the personalization of the information the use of the user profile has been adopted. This generates the problem of evolution of this profile over time. In this paper is proposed a model of evolution of the user profile based on the temporal Bayesian networks. This model is the result of a repetition over time on a sequence of T length, of a basic network representing a search activity characterized by the request submitted at t time and the user interest corresponding with the relevant documents and the terms of the index. The resulting networks are connected by causal links between each two successive steps of time, which allows the dynamic evolution of user profile indicated by the change of the user interests throughout the various submissions of user queries at the sequence of time