摘要
针对数据稀疏性问题,从提高稀疏数据矩阵利用效率这个角度,提出了一种基于间接评分的协同过滤算法,在基于用户和基于项目的协同过滤算法基础上,将2种算法的预测评分进行动态地混合加权作为直接预测评分,同时引入“相似用户”对“相似物品”的评分作为间接预测评分,最后把间接预测和直接预测2种评分加权形成用户对项目的最终评分.为证明该方法的有效性,使用MovieLens电影评分数据集对算法进行验证,结果表明该方法的平均绝对误差要比传统的基于用户和基于项目的协同过滤算法低,表明了在稀疏数据上该文提出的基于间接评分的协同过滤算法效果更佳.
Aiming at the problem of data sparsity,this paper proposes a collaborative filtering algorithm based on indirect scoring from the perspective of improving the efficiency of sparse data matrix.Based on the user-based and project-based collaborative filtering algorithm,the prediction scores of the two algorithms are used.Dynamically blending weights as direct predictive scores,while introducing“similar users”scores for“similar items”as indirect predictive scores,and finally weighting both indirect and direct predictions to form the user’s final score for the project.In order to prove the validity of the proposed method,the MovieLens film scoring dataset is used to verify the algorithm.The experimental results show that the average absolute error of the proposed method is lower than the traditional user-based and project-based collaborative filtering algorithm which indicates that the collaborative filtering algorithm based on indirect scoring proposed in this paper is more effective on sparse data.
作者
张超
颜伟
ZHANG Chao;YAN Wei(NetworkInformationCenter,QufuNormalUniversity,273165,Qufu,Shandong,PRC)
出处
《曲阜师范大学学报（自然科学版）》
CAS
2019年第3期60-65,共6页
Journal of Qufu Normal University(Natural Science)
关键词
数据稀疏性
间接评分
推荐系统
协同过滤
data scarcity
indirect rating
recommended system
collaborative filtering
