摘要
通过协同过滤获取用户的兴趣是为其提供贴心的个性化服务的关键技术.针对传统的协同过滤推荐算法不仅只考虑用户间单个特征的相似性,而且忽略用户兴趣会随时间变化而变化,从而难以准确地预测目标用户的兴趣,针对上述问题,提出一种基于多特征相似的用户兴趣推荐算法.在近邻居中寻找出与目标用户多特征相似的用户,根据相似用户对项目的评分以及目标用户兴趣随时间变化的时间函数来预测目标用户对该项目的评分,从而达到向目标用户推荐的目的.实验结果表明,该算法与传统的协同过滤推荐算法相比,能有效地提高用户推荐的质量.
Obtaining user＇s interests by collaborative filtering recommendation is the key technology to provide personalized service for the customer. The traditional collaborative filtering recommendation algorithm not only considers the similarity of users＇ single feature,but also ignores the change of users＇ interest with time. Thus, it is difficult to accurately predict the user interests. In view of the above problem,a recommendation algorithm of user＇s interest based on multiple similar features is proposed. First of all, users similar to the target user with multiple feature are found out in the neighborhood; then,in order to achieve the purpose of recommen- ding to the target user,the target user＇s assessment about the project is predicted based on the assessment of the similar users and the time function which represents change of the target user＇s interest with time. Experimental results show that the proposed algorithm is more effective than traditional collaborative filtering recommendation algorithm.
出处
《西安工程大学学报》
CAS
2016年第1期97-101,共5页
Journal of Xi’an Polytechnic University
基金
陕西省教育厅科学研究计划资助项目(14JK1307)
陕西省自然科学基金资助项目(2015JQ5157)
关键词
用户兴趣
多特征相似性
个性化推荐
协同过滤
时间函数
user＇s interests
multiple similar features
personalized recommendation
collabora-tive filtering
time function
