Sélection adaptative de Services de Recherche d'Information web par l'analyse du besoin et du comportement de l'utilisateur

As part of research on modeling user needs and behavior, we describe an approach to select web Information Retrieval Services (IRS) adapted to user’s needs. An experimental system integrating a model of the user by a user profile representing these interests, modeling the behavior by a mechanism for retrieving user interactions and a database of general and vertical SRI is presented. Our research focus on the construction of a selection model with features from the literature and the user profile. We propose the use of reinforcement learning using the theory of Markov decision process for learning our model. MOTS-CLÉS : Recherche d’Information, Modélisation de l’utilisateur, Selection de Services de Recherche d’Information