Resource recommendation based on topic model for educational system

In this paper, we propose a method for resource recommendation based on topic model in an e-learning system. The Web provides an extremely large and dynamic source of information. So it is now increasingly popular to provide personalized service in document recommendation. Personalized service can reduce information overload and, hence, increase user satisfaction. Topic model is a generative model for text mining, which has significant effects in both efficiency and accuracy. Latent Dirichlet Allocation (LDA), an approach to building topic models based on a formal generative model of documents, is and feasible and effective algorithm in text modeling. We propose an LDA-based interest model within the language modeling framework, and evaluate it on an e-learning system. In an e-learning system, topic model can provide a good vector model for the course document. Besides with the help of the topic model, we can build an exact model for users' interests, because in an e-learning system, we can get the users' access action and users' learning condition from the server. Thus the system can adopts interest mining technology and topic model to automatically identify the learner's interests and recommend interest-related resources to specific person. In this paper, we only focus on interests modeling and resource recommendation. The interest modeling system using proposed approach based on topic model is more effective. Meanwhile, the recommendation system based on user interests also gets better result.