E-Learning Course Recommendation Using Hybrid Fusion With Matrix Factorization

Abstract

Rapid development in Internet and communication technologies (ICT) has made E-learning as a rapidly adopted mode for delivering educational and training services. However with vast amount of contents, there is difficulty for a learner to search for the contents of interest and there is need for content personalization. Content recommendation is an important part of content personalization service in E-Learning, where the course contents are recommended to user based on multiple parameters like users content viewing history, user profile, similar user content viewing history, content popularity etc. In this paper, we develop a hybrid fusion with matrix factorization for content recommendation in E-Learning environment. The system provides recommendation based on multiple dimensions of user and content profiles.  Different from the previous works which relies only on rating, this work models the user interest on content and content interrelation in terms k-gram concepts. Due to this concept modeling, the recommendation accuracy is higher than previous approaches.