基于用户标注行为的相关性分析及重排序  

Relativity Analysis and Re-ranking Based on User Annotation Behavior

作　　者：李枫林[1] 张景[1] 

机构地区：[1]武汉大学信息管理学院,湖北武汉430072

出　　处：《情报理论与实践》2010年第10期57-61,共5页Information Studies:Theory & Application

基　　金：教育部人文社会科学重点研究基地重大项目(项目编号:07JJD870220)和教育部人文社会科学一般项目(项目编号:07JA870009)资助

摘　　要：用户标注行为反映了标注对象与标注结果之间的相关关系。本文通过对用户标注行为的分析,详细研究了用户标注行为所反映的网页间相关性、标签间相关性以及网页和标签间相关性的关联程度,并将这种相关性分析用于标签的相关性计算上,改进了SPR算法。结果表明该算法可以有效提高检索结果重排的效果。User annotation behavior reflects the relationships between annotated objects and tags. Based on the analysis of the user annotation behavior,this paper makes a detailed study of the relativity between Web pages,the relativity between tags and the relationships between them reflected by the user annotation behavior. The paper applies the relativity analysis to the relativity computation of tags to improve SPR. The results show that the algorithm can improve the re-ranking effect of retrieval results effectively.

关 键 词：用户标注 标签 算法 

分 类 号：G354[文化科学—情报学]