Personalized Learning Resource Recommendation Based On Learner Profile

The online learning platform is overloaded with learning resources, and learners are prone to “information maze” when facing massive learning resources. An effective way to solve information maze is to recommend resources that conform to learners’ own characteristics from a large number of learning resources. Therefore, we proposes a personalized learning resource recommendation algorithm based on learner profile. Firstly, the learner portrait is modeled based on two characteristics of learner, learning interest preference and knowledge point mastery level. Secondly, K nearest neighbor learners who are most similar to the target learner are found according to the learner portrait. Thirdly, using the rating information of these K learners on the learning resources, the predicted scores of the target learner on the learning resources are calculated. Finally, the learning resources are sorted by predicted scores, the first N learning resources are the final recommended results. The algorithm we proposed is verified in EdNet, a data set published by Santa. The experimental results show that the algorithm has significantly improved the recommendation performance.