摘要
协同过滤技术是目前电子商务推荐系统最为有效的信息过滤技术之一.最近的研究尝试在推荐过程中引入信任模型来提高推荐的准确性和抵御"托"攻击.但在用户多兴趣的情况下,属于不同主题的项目需要不同的可信赖人员来推荐,传统的概貌级信任模型已不再适用.本文提出主题级信任计算模型以及基于主题级信任的协同过滤算法.一系列的实验结果表明,该算法在不牺牲鲁棒性的同时,有效地提高了推荐的准确性.
Collaborative filtering technique is one of the most effective information filtering techniques in E-commerce recommender systems to date. Recent researches try to incorporate trust model into recommendation process to improve the accuracy and resist shilling attack. However, because of user ts multiple interests, items belonging to different topics need different trustworthy users to make recommendation. Traditional profile level trust model is not suitable. Based on this idea, this paper proposes a topic-level trust model and topic-level trust based Collaborative filtering algorithm. A series of experiments show that the algorithm achieves highly on the accuracy and robustness.
出处
《小型微型计算机系统》
CSCD
北大核心
2008年第8期1415-1419,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(70663002)资助
江西省教育科学"十一五"规划课题(06YB141)资助
江西省社会科学"十一五"规划项目(06JY13)资助
关键词
电子商务
推荐系统
协同过滤
“托”攻击
信任
e-commerce
recommender system
collaborative filtering
shilling attack
trust
