结合用户信任模型的协同过滤推荐方法研究  

Research of collaborative filtering recommendation based on user trust model

作　　者：蔡浩[1] 贾宇波[1] 黄成伟[1] 

机构地区：[1]浙江理工大学信息电子学院,杭州310018

出　　处：《计算机工程与应用》2010年第35期148-151,共4页Computer Engineering and Applications

基　　金：浙江省教育厅科研计划项目(No.20060599);浙江理工大学科学基金项目(No.111251A4Y04002)

摘　　要：协同过滤推荐是当前最成功的推荐技术之一,在电子商务推荐服务中得到了广泛的应用,它根据和目标用户具有相似行为的用户对项目的评价来进行推荐。鉴于传统的协同过滤推荐算法过于强调相似性的作用,并且和用户的认知习惯矛盾,引入了社会学中较成熟的信任机制来改进传统算法。实验结果表明,改进方法是有效的,它和传统的协同过滤推荐算法相比有更好的推荐质量。Collaborative filtering is one of the most successful recommendation technology,which has been widely used in e-commerce recommendation,and it uses the ratings of users who have similar behavior with target user to generate recom-mendation.Howeverc,urrent research reveals that the traditional collaborative filtering algorithms emphasize on the role of sim-ilarity too much,which is a contrary to our cognition.In this paper,we introduce the mechanism of trust which is mature in sociology to improve the traditional algorithm.The experiment result shows that the improvement algorithm is efficient since it has higher accuracy compared with the traditional collaborative filtering.

关 键 词：个性化推荐 协同过滤 信任度 用户信任模型 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]