一种考虑双重相似度的协同过滤推荐算法

协同过滤算法为推荐系统提供了一种方法,但传统的协同过滤方法推荐精度低.提出一种考虑用户评分相似性的协同过滤算法,通过在皮尔逊相关系数中加入项目数量相似度和用户评分相似度两个因素来计算用户间的相似度,以产生更合理的邻居用户,提高推荐精度,完成对用户的推荐,同时邻居用户的选取采用动态阈值设定方法.实验结果表明,所提出的算法相比传统方法选择出的邻居更为精确,推荐质量更高.

A Collaborative Filtering Recommendation Algorithm Considering Double Similarity

Collaborative filtering algorithm provides a way to recommendation systems,however,the precise of these methods used in traditional collaborative filtering recommendation is low.In this paper,a collaborative filtering recommendation algorithm considering double rating similarity was proposed,adding two elements which were the number of same user recommendation and the user rating similarity,so as to provide the more reasonable neighbor users,enhance the recommendation precision,complete the recommendation to user,and the dynamic threshold method to choose the neighbor users was used.The results shows that the proposed algorithm is more precise compared with the traditional method in choosing neighbor users,and the recommendation has the higher quality.