The increasing of multimedia information great diffusion on the web generates cognitive users' overload and plenty of solutions are proposed to solve it. Despite the multitude of solutions, web users are still looking for more and simpler assistant in web using. In this paper, we present an information adaptation system using characteristics of the user. Our work aims to improve access to information through recommendations made on the basis of relevance notices (feedbacks) given by other users of the system when they access information. After introducing the concepts and mechanisms used by our system, we compare it with other systems in order to show the originality of our ideas. Finally, we propose an implementation and a preliminary evaluation of our software in order to assess its performance.