基于叙词表语义关系和PageRank的查询扩展方法  

An Approach for Query Expansion Based on Semantic Relation of Thesaurus and Page Rank

作　　者：何伟[1,2] 常春[1] 

机构地区：[1]中国科学技术信息研究所,北京100038 [2]怀化学院,怀化418008

出　　处：《情报杂志》2016年第12期105-110,共6页Journal of Intelligence

基　　金：中国博士后科学基金项目"基于叙词表语义关系的智能检索模型研究"(编号:2014M550791)研究成果之一

摘　　要：[目的/意义]查询扩展是一种有效的提高信息检索系统查全率和查准率的方法,一直以来受到大家的关注。[方法/过程]提出一种基于叙词表语义关系和PageRank的查询扩展方法,首先通过叙词表为原始查询词获得初始查询扩展词集,并根据叙词表词间语义关系构造初始查询扩展概念树;其次使用修正的PageRank方法 Modified PageRank计算初始查询扩展概念树中每个节点的重要性M PR值,通过经验阈值对其进行剪枝操作;获得最终的查询扩展词集。[结果/结论]实验结果证明,这一方法能进一步改善信息检索系统的查全率和查准率,平均Fmeasure值有一定幅度的提升,比无扩展方法提高了5.59%,表现出较强的竞争力。[ Purpose/Significance ] Query expansion is a method for improving the index of recall and precision of information retrieval, and has always been the focus of many researchers. [Method/Process]In this paper, we propose an approach for query expansion based on semantic relation of thesaurus and PageRank. At first, it obtains an initial query expansion set for an original query word using thesaurus, and constructs an initial conceptual tree of query expansion based on the semantic relation between words in thesaurus; secondly, we exploit Modified PageRank to measure the importance MPR of each node in initial conceptual tree of query expansion, and prune the conceptual tree by an empirical threshold; finally, a final word set of query expansion is obtained. [ Result/Conclusion] Experimental results show that the method proposed in this paper can improve the index of recall and precision of information retrieval, and average F-measure is promoted in a certain scope and increases 5.59% in contrast to the method of no query expansion, the method shows great competiveness.

关 键 词：语义关系 PAGERANK 查询扩展 叙词表 概念树 

分 类 号：G254[文化科学—图书馆学]