支持Web2.0标签层次体系构建的关系识别及层次组合方法研究  

Research on Relation Recognition and Hierarchy Composition for Web 2.0 Tag Hierarchy Construction

作　　者：高克宁[1] 张引[2] 张斌[2] 张聿博 

机构地区：[1]东北大学计算中心,辽宁沈阳110004 [2]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《电子学报》2014年第1期58-61,共4页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61073062);辽宁省科学技术基金(No.20102060);沈阳市科学技术计划(No.F11-264-1-33)

摘　　要：从标签系统中生成层次体系可以支持多种类型的应用,具备重要的意义.当前的研究主要集中于发现标签间的关系,但对如何利用这些关系形成高质量的层次体系却关注不足.针对这一现状,研究了支持Web 2.0标签层次体系构建的关系识别及层次组合方法,通过分析并识别已发现的标签间关系所具有的不同类型提升了标签间关系的质量,并提出基于语义流动分析的层次组合方法实现了更高质量的层次体系构建.应用多种评估指标的实验结果表明应用关系识别及使用语义流分析方法可以获得相比评估方法更高质量的标签层次体系.Tag hierarchies generated from social tagging systems can be used in various fields. Current research works mainly focus on the detection of tag relations, but pay much less attention on how to generate high quality tag hierarchy with these rela- tions. This paper studies relation recognition and hierarchy composition methods for Web 2.0 tag hierarchy construction. By analyz- ing and recognizing the types of the tag relation detected by available methods, quality of tag relation is improved. Moreover, a se- mantic flow analysis based hierarchy composition method is proposed to acquire tag hierarchy with higher quality. Experimental re- sults measured by various evaluation metrics have shown that by applying relation recognition and semantic flow analysis based hierarchy composition,hierarchies with higher quality with respect to other evaluated methods can be yielded.

关 键 词：WEB2 0 社会标注 标签层次体系 关系识别 层次组合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]