Nowadays, the online interaction has become more necessary and vital in terms of both socio-cognitive and socio-affective levels. It fully plays an important role in the acquisition of knowledge in online learning. However, the way in which these interactions take place remain little regulated and less efficient. Recognizing this, we have developed an approach used to automatically schedule the interactions in the learning process of learners. These interactions will easily help learners to assimilate the obscure concepts and quite difficult to understand. Our goal is achieved basing on intelligent agents modeling of virtual learners. These agents enter into discussion with learners and are widely involved in clarifying these concepts. To affirm the importance of our study in e-learning, we have given a questionnaire to the learners of the Master in French literature. This latter has confirmed that the majority of the learners choose to collaborate and interact with each other to come over the blockage points in their learning process. Furthermore, the empirical result has shown that our approach contributes effectively in online learning. A significant increase in learner results is well noticed.