Research on information recommendation method of online education platform based on user preference model

Abstract

At present, there are few personalized information recommendation services in online education platforms. The resources received by users are basically the same, and there is no personalized recommendation based on user preferences. How to accurately locate relevant resources from massive resources according to the needs of learners, and provide personalized online education services for them, has become a current research hotspot. Although the research on recommendation algorithms in other fields has achieved good results, there are relatively few applications of online course resource recommendation systems. This paper improves the collaborative filtering recommendation algorithm. On the basis of the traditional algorithm, the dual-attribute scoring matrix is used for attribute division and the BP neural network is used for scoring prediction. This improvement is to fill the scoring matrix and solve the problem of the recommendation quality degradation caused by "cold start" and too sparse scoring data in traditional algorithms. Finally, the effectiveness of the improved algorithm is proved by experiments.