摘要
为了提高传统协同过滤推荐算法推荐的准确度,对评分信任和社交信任赋予自适应的权重,结合概率矩阵分解算法,提出一种综合的个性化推荐算法.该算法在Filmtrust数据集上进行验证,并与相关算法进行对比,结果表明所提算法在MAE(mean absolute error)和RMSE(root mean squared error)指标上均得到有效的改进.
By giving the adaptive weight to the user's trust in project score data and social trust,a comprehensive personalized recommendation algorithm combining multi-view trust and probability matrix decomposition was proposed to improve the recommendation accuracy.The algorithm was verified on the Filmtrust dataset.The results showed that the proposed algorithm was effectively improved on both MAE(mean absolute error)and RMSE(root mean squared error)indicators compared with related algorithms.
作者
任志波
王亚文
魏翔宇
管成康
REN Zhibo;WANG Yawen;WEI Xiangyu;GUAN Chengkang(Liberal Arts Comprehensive Experimental Center,Hebei University,Baoding 071002,China;College of Management,Hebei University,Baoding 071002,China;School of Humanities and Sciences,Northeast Petroleum University,Daqing 163318,China;School of Information Science and Technology,Xiamen University Malaysia,Sepang 43900,Selangor Darul Ehsan,Malaysia)
出处
《河北大学学报（自然科学版）》
CAS
北大核心
2020年第5期552-560,共9页
Journal of Hebei University(Natural Science Edition)
基金
国家社会科学基金资助项目(17BJY116)
河北省社会科学基金资助项目(HB19TQ011)。
关键词
信任
概率矩阵分解
协同过滤
个性化推荐
trust
probabilistic matrix factorization
collaborative filtering
personalized recommendation
