作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 开发研究与工程应用 • 上一篇    下一篇

基于关键词语义与作用域扩展的事件检测

褚衍杰,魏 强,李云照   

  1. (盲信号处理重点实验室,成都 610041)
  • 收稿日期:2013-07-09 出版日期:2014-08-15 发布日期:2014-08-15
  • 作者简介:褚衍杰(1982-),男,博士研究生,主研方向:信息处理;魏强,硕士;李云照,博士。

Event Detection Based on Keyword Semantics and Scope Extension

CHU Yan-jie,WEI Qiang,LI Yu-zhao   

  1. (Key Laboratory of Blind Signal Processing,Chengdu 610041,China)
  • Received:2013-07-09 Online:2014-08-15 Published:2014-08-15

摘要: 针对具有关联要素的中文文本事件检测问题,提出一种新的事件检测算法。采用关联关系分析、关键词语义扩展和关键词作用域扩展等方法,将目标事件包含的所有文本作为一个整体进行关键词匹配。该算法将关键词向量分为2个子向量,依次利用关键词子向量对待检测文本进行关键词语义扩展、关键词匹配和关联关系分析,形成目标事件的文本集并提取事件信息。实验结果显示,在关键词较少的条件下,该算法通过关键词语义扩展提高检测到目标事件的概率,在关键词较多时,通过关键词作用域扩展提高检测到目标事件的概率,可降低漏检率。

关键词: 事件检测, 语义扩展, 作用域扩展, 关联关系, 信息提取, 事件模板

Abstract: A new event detection algorithm is proposed to solve the problem of event detection of texts with association relationship.The algorithm uses the method of association relation analysis,semantics and scope extension of the keywords.In this way,all texts in an event are regarded as one text to match the keywords.The algorithm separates the keywords into two groups,then executes keyword semantics extension,keyword matching and association relationship analysis with the groups to get all texts in the event,and extracts and affirms important information.Experimental results indicate that when the keywords are less,the algorithm can improve the probability of getting the event using semantics extension of the keywords,and when it has a large number of keywords,the algorithm can improve the detection rate and reduces the false negative rate using scope extension of the keywords.

Key words: event detection, semantics extension, scope extension, association relation, information extraction, event template

中图分类号: