基于用户特征和用户兴趣变化的协同过滤推荐  

Collabrative Filtering Recommendation Based on Users＇ Feature and Users＇ Drifting Interest

作　　者：吴婷[1] 熊前兴[1] 贺曦春[1] WU Ting, XIONG Qian-xing, HE Xi-chun （Department of Computer Science and Technology, Wuhan University of Technology, Wuhan 430063, China）

机构地区：[1]武汉理工大学计算机科学与技术,湖北武汉430063

出　　处：《电脑知识与技术》2008年第12期1730-1732,共3页Computer Knowledge and Technology

基　　金：长江航道信息化专项项目（20071h0347）

摘　　要：协同过滤系统是目前最成功的一种推荐系统，但是传统的协同过滤算法没有考虑用户兴趣会随时间发生变化以及类似特征用户对用户相似度精度具有影响等因素，导致推荐质量较差。该文结合用户兴趣变化和用户特征两个因素，提出了新的用户之间相似度计算方法用来提高协同过滤推荐质量。Collaborative filtering （CF） is the most successful recommended system to date, but traditional CF algorithm does not consider the problem of drifting users＇ interests and users＇ feature which often results in poor recommendation. This paper combined two factors of drifting users＇ interests and user＇ feature. A new method of computing similarities between different users is developed which can enhance accurate of commendation.

关 键 词：协同过滤 兴趣变化 用户特征 

分 类 号：TP393[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]