Abstract EN:

Access to relevant information adapted to user’s preferences and contexts is a challenge in many applications. In this thesis, we address the personalization in the context of content delivery platforms. Hence we propose a personalized access model based on multidimensional models of user’s profile and context. The PAM provides a set of services that enable applications to take into account both user’s profile and context within personalization processes, hence delivering more accurate contents. PAM services include, but are not limited to, an automatic approach for contexts and contextual preferences discovery, the projection of user’s profile within his current context, and matching of profiles and contents to provide user recommendations. We also show that PAM services allow a smooth integration of context within personalized applications without changing their inner processes. Thus, we instantiated the PAM to define context-aware recommender systems used to evaluate our approach.