基于重要性排名的协同过滤推荐算法

摘要

Traditional collaborative filtering completely neglects the importance of the items and users,which will affect the recommendation quality.So,a new algorithm based on the importance rank is proposed,in which the importance of items is integrated into user similarity measure method and importance of users is integrated into score prediction computing method.Besides,to measure the importance of items and users in the system,ItemRank and UserRank algorithms are proposed.The results of the experiment on the MovieLens dataset show that the proposed algorithm improves the recommendation quality efficiently.%传统的协同过滤忽略系统中不同用户和条目的重要性对推荐结果的影响.针对此问题,提出了一种基于用户和条目重要性的改进协同过滤算法,该算法将条目的重要性融合到用户相似性的度量方法中,将用户的重要性融入到预测评分的计算方法中；为度量系统中每个条目和用户的重要性,提出了ItemRank和UserRank算法.在MovieLens数据集上的实验结果表明,提出的算法可以显著提高推荐系统的推荐质量.