基于查询词邻近度的专家搜索算法  

Expert Search Algorithm Based on Query Word Proximity

作　　者：杨柳[1] 张文生[1] 

机构地区：[1]中国科学院自动化研究所复杂系统与智能科学重点实验室,北京100190

出　　处：《计算机工程》2011年第6期7-8,11,共3页Computer Engineering

基　　金：国家自然科学基金资助项目(90924026);国家"863"计划基金资助项目(2008AA01Z121;2007AA01Z338)

摘　　要：提出一种基于查询词邻近度的专家搜索算法。根据查询词在窗口中的共现关系构建概率语言模型的查询词邻近度,利用经典概率模型作为背景平滑,通过对多个查询词在文档中的距离关系建模,结合候选人与查询词之间的距离对候选专家排序。实验结果表明,在该算法中引入查询词邻近度可以提高搜索准确率,应用候选人与查询词2种邻近度可以取得更好的专家搜索效果。This paper proposes an expert search algorithm based on query proximity.It builds query word proximity of probability language model according to query word concordance in window,and uses classical probability model as background smooth.The distance dependencies between multiple query terms and it then ranks the candidates and combines with the distance between candidates and query terms.Experimental results show that employing query proximity in expert search improves the precision score significantly and a combination of both kinds of dependencies can achieve better performance.

关 键 词：专家搜索 查询词邻近度 概率语言模型 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]