摘要
通过用户行为信息并结合信任传递推断用户隐式信任关系,提出了基于矩阵分解的PTtrustSVD算法,并在Filmtrust数据集上进行了实验.结果表明,加入隐式信任关系优于仅使用显式信任关系的推荐方法,证明了隐式信任关系对于改进推荐系统性能的有效性.
Based on the similarity of user behavior information and trust transferring to infer the implicit trust relationship, this paper proposes a novel algorithm that integrated with matrix composition technique, namely, PTtrustSVD. The experimental results show that the purposed method using the implicit trust relation outperforms that only using explicit trust relation in the Filmtrust dataset, and prove that the implicit trust relation is effective for the recommender system.
出处
《东北师大学报(自然科学版)》
CAS
CSCD
北大核心
2017年第4期71-75,共5页
Journal of Northeast Normal University(Natural Science Edition)
基金
国家自然科学基金资助项目(71473035
11501095)
吉林省科技发展计划项目(20150204040GX
20170520051JH)
吉林省发改委项目(2015Y055)
东北师范大学自然科学基金资助项目(2014015KJ004)
关键词
推荐系统
隐式信任
信任传递
recommender system
implicit trust
trust transferring