关键词: 推荐系统, 社交网络, 协同过滤, 用户评分偏好, 评分预测

Abstract: To solve the severe sparseness problem of traditional collaborative filtering recommendation algorithm, this paper proposes a novel collaborative filtering recommendation algorithm based on the characteristics of social network. On the basis of traditional matrix decomposition model, the algorithm obtains the trust and trusted characteristic matrix by integrating the characteristics of social network and user??s preference degree, and then, predicts the rating of the commodity by the social identity matrix, the commodity characteristic matrix and the user rating preference similarity in common. In order to verify the reliability of the proposed algorithm, this paper uses the Epinions open dataset to compare the algorithm performance. The experimental results show that compared with the existing social recommendation algorithms, the proposed algorithm has smaller average absolute error and root mean square error. Meanwhile, there is a linear relationship between the time complexity of the proposed algorithm and the number of the dataset. Therefore, the proposed algorithm can effectively reduce the impact of data sparseness on recommendation results and improve the recommendation accuracy rate. In practice, the proposed algorithm can be considered as an alternative and development of the large-scale data set recommendation.

Key words: recommender system, social network, collaborative filtering, user rating preference, rating prediction