基于关联词对动态抽取的报道关系检测技术研究  

Story Link Detection Research Based on the Dynamic Extraction of Correlative Word

作　　者：赵华[1] 邓攀[2] 张建伟[2] 

机构地区：[1]山东科技大学信息科学与工程学院,青岛266510 [2]北京航空航天大学软件开发环境国家重点实验室,北京100191

出　　处：《计算机科学》2010年第6期237-239,270,共4页Computer Science

基　　金：国家自然科学基金(60773034);山东科技发展计划项目(2008GG30001024);山东科技大学"春蕾计划"项目资助

摘　　要：报道关系检测是判断随机选取的两个新闻报道是否讨论同一话题的技术。提出了一种基于关联词对动态抽取的报道关系检测方法。关联词对是指在同一篇报道中出现的满足一定关系约束的两个单词,而关系约束是指一组特征的集合。该方法认为两篇报道中出现的相同的关联词对越多,两篇报道的相似度越大。实验证明基于关联词对动态抽取的报道关系检测方法取得了非常好的效果,从而证实了所提方法的有效性。同时,实验还表明,关系约束对该方法的成功实施起着非常重要的作用。Story link detection is a task of topic detection and tracking, which is to detect whether two stories are ＂linked＂, or discuss the same event. This paper proposed a story link detection method based on the dynamic extraction of correlative word. Correlative word is defined to be a pair of words that satisfy certain Relation Restriction. In this paper, relation restriction refers to a set of features. This paper explored three features, which include the initial letter, location and window. The more two stories had the same correlative words, the larger similarity they bad. Experimental results showed that the story link detection method based on the dynamic extraction of correlative word performs very well,which testifies the great capabilities of this method. We also find that relation restriction is critical to the performance of this method.

关 键 词：话题检测与跟踪 报道关系检测 关联词对 关系约束 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]