Differential Privacy Protection Recommendation Algorithm Based on Student Learning Behavior

Abstract

Traditional collaborative filtering recommendation algorithm based on learning resources use a large amount of student personal information and behavior information. This will put the user's privacy at risks since that students' information can be mined by analyzing the recommendation results. Considering that differential privacy theory can effectively protect user privacy through strict mathematical definition and maximum background knowledge assumptions, this paper proposes a differential privacy collaborative filtering recommendation algorithm based on learner behavior similarity. By adding noise obeying the Laplace distribution to the learner behavior similarity matrix, the recommendation accuracy rate does not reduce, as well as the privacy of student is protected effectively.