Abstract:

Nowadays the amount of educational content is increasing every day. Recommender systems are used in e-learning platforms to provide the learners with specific suggestions and relevant resources. However, one of the main issues that are faced on the e-learning recommendation environment is the "cold start problem", the System is unable to report relevant content due to a lack of information about learners at the beginning of the use. In this paper, an analysis of some existing architectures is conducted by presenting their contribution and criticizing the aspects they missed. The new approach of the recommender e- learning system based on the collaborative filtering to minimize the cold start problem for the user’s suggestions is proposed. This take into account the three stages of the cold start compared to other reported contributions.