基于社会化标注的用户兴趣挖掘  

User preference mining based on social tagging

作　　者：扈维[1] 张尧学[1] 周悦芝[1] 

机构地区：[1]清华大学计算机科学与技术系,信息科学与技术国家实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2014年第4期502-507,共6页Journal of Tsinghua University(Science and Technology)

基　　金：国家"八六三"高技术项目(2011AA01A203);国家科技支撑计划项目(2012BAH13F04)

摘　　要：用户兴趣挖掘是实现个性化推荐与智能化服务的关键问题。Web2.0引入的社会化标注可以反映用户的潜在兴趣。该文提出一种基于用户标注行为的兴趣建模方法,根据用户与标签的交互模式反映用户的兴趣倾向。从用户对不同标签的"认同度"和"依赖度"两方面衡量用户的标签兴趣,并使用"标签基因"对用户的兴趣进行细粒度分解。来自真实用户数据的实验结果表明,该方法可以有效提高用户兴趣的预测准确度和覆盖率,创建的兴趣模型更加符合用户的真实情况。User preference mining is one of the key problems in personalized recommendations and intelligent services. Social tagging in web2.0 reflects the user＇s potential interests. This paper presents a user preference modeling method based on social tagging that predicts user preferences based on interactions between user and tag. The user＇s ＂degree of recognition＂ and ＂dependency＂ on an individual tag are combined to evaluate the user＇s tag preference. The user＇s interest is then decomposed into a fine-grained result using a ＂Tag Genome＂. Tests based on real data demonstrate that this method significantly improves prediction accuracies and coverage to more accurately match the user＇s real interests.

关 键 词：用户模型 社会标注 兴趣挖掘 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]