In distance learning, recommendation system (RS) aims to generate personalized recommendations to learners, which allows them an easy access to various contents at any time. This paper discusses the main RSs employed in E-learning and identifies new research directions to overcome their weaknesses. Existing RSs such as content-based, collaborative filtering-based and knowledge-based recommendations reveal significant softness due to their incapacity to collect accurate information about learners, especially new ones which is identified as cold start problem. In this paper, we are working on both, the new user cold start problem, which is considered as a big issue in E-learning system, and the recommendation based on updating information. This complication can be reduced by including other learners' information in the process of recommendation. The objective of this study is to propose an E-learning Recommender System based on Dynamic Ontology. Our recommended approach describes semantically course and learner, which will be integrated into Collaborative and content-based filtering techniques to generate the top N recommendations using clustering methods. The experiment's measures are done using the famous "COURSERA" dataset mixed to our university USMBA dataset. The results obtained demonstrate the effectiveness of our proposed method in the process of recommendation compared to content-based method.