Towards an Adaptive E-learning System Based on Q-Learning Algorithm
Mohamed Boussakssou, Bader Hssina, Mohammed Erittali
Abstract
The challenge for today’s online learning systems is to provide effective access to knowledge and content that is relevant to learners’ expectations. The majority of these systems lack methods to support the needs of learners who are generally heterogeneous in terms of intellectual abilities, learning pace, preferences, etc. There is a need to provide powerful mechanisms to organize such learning and to adapt pedagogical decisions to the particular skills and needs of each learner. Our contribution in this area of research is the development of an adaptive e-learning system that can generate learning paths adapted to the profile of the learner. Indeed, we propose an approach to dynamically compose adaptive online learning courses based on learner activities, learning objectives, and instructional design strategies using the Q-learning algorithm, which is a reinforcement learning technique. The latter is based on the behavior of the learners and provides the course content necessary to achieve the learning objectives according to the positive and or negative feedback of the learners. In addition, we conclude with experience and evaluation of our approach based on the Q-learning algorithm.