关键词: 协同过滤，时间效应，用户偏好度，用户特征向量, 协同过滤,推荐系统,共同评分项,结构相似性,电子商务

Abstract: The collaborative filtering recommendation algorithm is one of the most important recommendation technologies in E-commerce recommendation system,and the similarity measuring method plays a key role for the accuracy of recommendation results.However,the traditional similarity measure methods ignore the influence on recommendation quality resulting from the number of the common grading items between users.Given this situation,a novel approach was firstly proposed based on the number of the common grading items when measuring the similarity between users.Further more,to protect recommendation result from the data sparsity,the structural similarity measure method of complex network was employed to evaluate the similarity between users.The experimental results show that the proposed approaches can avoid the disadvantages of traditional methods effectively and improve the quality of the recommendation.

Key words: Collaborative filtering, Time effect, Uscr preference degree, Uscr characteristic, Collaborative filtering,Recommendation system,Common grading items,Structural similarity,E-commerce