基于查询词相关性传播的专家检索方法.pdfVIP

2014 年 6 月 计算机工程与设计 June 2014 COMPUTER ENGINEERING AND DESIGN Vo l. 35 No.6 第 35 卷第6 期 基于查询词相关性传播的专家检索方法 郑义平，王勇，李佳，王瑛 (广东工业大学计算机学院，广东广州 510006) 摘 要:针对专家检索存在性能低等问题，提出一种基于查询词相关性传播的专家检索方法，不仅考虑与专家相关的文档 内容，还考虑了专家在文档中的共现关系。将用户输入的查询切分成多个查询词，使用经典的专家语言模型计算专家关于 查询坷的初始得分，根据专家的共现关系构建带权有向图，采用相关性传播模型对初始得分进行提升，计算得到专家关于 查询的最终得分。实验结采表明，该方法对专家检索性能有很好的提升。 关键词:专家检索;查询词;专家语言模型;共现关系;相关性传播 中圄法分类号: TP391 文献标识号:A 文章编号: 1000-7024 (2014) 06-2165-05 Expert finding method based on term relevance propagation ZHENG Yi-ping , WANG Yong , LI Jia , WANG Ying (School of Computer , Guangdong University of Technology , Guangzhou 510006 , China) Abstract: An expert finding method based on ten丑 relevance propagation was proposed to solve the low retrieval perfo口nance problem in expert retrieva l. This method took not only the document content , but also the cσoccurrence relationship between experts in the document into consideration. Firstly the query was segmented into several terms , terms initial expertise score was calculated by using expert language model , and then weighted directed graphs were constructed according to the co-occurrence re- lationship , and the initial expertise scores were promoted by relevance propagation mode l. Finally the query scores were gotten. The experimental results showed that this method promoted the expert retrieval performance. Key words: expert retrieval; term; expert language model; ccγoccurrence relationship; relevance propagation 些共同知识的特点，提出一种基于组织层次结构的专家检 。寻|言 索方法。另外Deng 等人[7J 将社区概念引入到专家检索中， 把社区作为专家与查询词之间关联的桥梁进行建模。 近年来，关于专家检索方法的研究已经提出了