Abstract de la publi numéro 5722

Both user modeling and multimedia adaptation communities have been actively carrying out research on adaptation, but the design of adaptation decision-taking engines is still an open issue for each application. In this paper we consider the decision-taking engines by using machine learning methods and propose an experimental adaptation platform. We also briefly describe actual implementations of this platform with 3 case-studies: supervised learning for teaching systems, reinforcement learning for multimedia streaming and unsupervised learning for adaptive hypermedia navigation support.