The MOOCs (Massive Open Online Courses) are durably modifying the learning methods in initial trainings. However, the registered learners in a MOOC are very heterogeneous. Moreover, their initial motivations and profiles may vary. The high dropout rate by learners shows that the concept of proposing a sequential or poorly structured learning path is not attractive enough. To address this problem, we propose in this work, an approach favoring the attractiveness of learners based on an adaptive approach within MOOCs. The federating element of our approach is defined on an overall learner model which aggregates its competencies and practices in social networks. Our proposals are implemented in a new platform MOOC: Claroline Connect.