Collaborative Filtering Algorithm Based on Improved Similarity Calculation
Zhihe Wang, Suping Shi, Hui Du, Shuyan Wang · 2019
With the increasing contradiction between the vast supply of information and the diversity needs of users, it is a wise choice to apply personalized recommendation to relevant areas. Among them, collaborative filtering algorithm is the basic algorithm in personalized recommendation, and similarity calculation is the core of the algorithm. The accuracy of its calculation has a great impact on the recommendation results. In order to solve the problem that the traditional similarity calculation results have great deviation and the recommended results are not ideal, a new similarity calculation method is proposed in this paper. According to the closer the rating time is, the higher the user's similarity is, and the time factor is incorporated into the proposed algorithm. At the same time, in order to avoid the improper contribution of active users and popular item to the similarity calculation, the penalty factor is added respectively. The experiment was carried out on MovieLens data set, and the experimental result showed that the proposed similarity method improved the accuracy of scoring prediction to some extent, and the recommendation effect was better than the traditional one.