修正评分的协同过滤算法

Modified Score Collaborative Filtering Algorithm

针对传统协同过滤算法推荐效率低、矩阵稀疏、用户异常值等问题,本文提出修正评分的协同过滤算法.算法先对用户聚类提高最近邻搜索效率,降低搜索时间.与此同时为提高算法推荐精度以及避免用户异常值问题,本文结合置信度提出修正评分代替传统算法中的加权评分衡量推荐结果.经验证,修正评分的协同过滤算法提高了推荐的准确性、有效性.

Low efficiency,matrix sparseness,user outliers,etc.for traditional collaborative filtering algorithms,this paper proposes a collaborative filtering algorithm for modified scores.The algorithm first clusters users to improve the efficiency of nearest neighbor search and reduce search time.At the same time,in order to improve the accuracy of algorithm recommendation and avoid the problem of user outliers,this paper proposes a modified score in combination with the confidence score to replace the weighted score in the traditional algorithm to measure the recommendation result.It has been verified that the collaborative filtering algorithm of the revised score improves the accuracy and effectiveness of the recommendation.