摘要
为了解决评分数据的稀疏性和用户最近邻的精确性问题,文章提出了一种基于奇异值分解(singular value decomposition,SVD)和项目属性的协同过滤推荐算法。该算法首先采用SVD方法对用户-项目评分矩阵降维,得到用户矩阵和项目矩阵,根据项目矩阵计算项目间的评分相似度,同时根据项目属性计算项目间的属性相似度,将2种相似度的结果加权计算得到项目间的相似度,最后采用最近邻的方法预测目标用户对待评分项目的评分。在MovieLens数据集上的实验结果表明,该文所提出的方法可以有效应对用户评分稀疏的问题,并能提高推荐的准确性。
In order to solve the problems of the sparsity of the scoring data and the accuracy of the user nearest neighbor,a collaborative filtering recommendation algorithm based on the singular value decomposition（SVD）and item attributes is proposed.Firstly,the SVD method is used to reduce the dimension of the user-item scoring matrix.Then the user matrix and the item matrix are obtained.The similarity between the items is calculated according to the scoring similarity between the items obtained by the item matrix and the attribute similarity by item attributes.Finally,the nearest neighbor method is used to predict the score of the target user to the target item.The results of the experiment on MovieLens dataset show that the proposed method can effectively solve the problem of data sparsity,and improve the accuracy of the recommendation.
作者
张建军
陆国生
刘征宇
ZHANG Jianjun;LU Guosheng;LIU Zhengyu(School of Computer and Information,Hefei University of Technology,Hefei 230009,China;School of Mechanical Engineering,Hefei University of Technology,Hefei 230009,China)
出处
《合肥工业大学学报（自然科学版）》
CAS
北大核心
2018年第6期761-765,858,共6页
Journal of Hefei University of Technology：Natural Science
基金
国家国际科技合作专项资助项目(2015DFI12950)
关键词
协同过滤
项目属性
奇异值分解(SVD)算法
数据稀疏
综合相似度
collaborative filtering
item attribute
singular value decomposition（SVD） method
data sparsity
comprehensive similarity
