检索
E-mail
RSS
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 ›› 2012, Vol. 38 ›› Issue (04): 37-39.doi: 10.3969/j.issn.1000-3428.2012.04.012

• 软件技术与数据库 • 上一篇    下一篇

基于最大熵的句内时间关系识别

王风娥，谭红叶，钱揖丽   

  1. (山西大学计算机与信息技术学院，太原 030006)
  • 收稿日期:2011-08-17 出版日期:2012-02-20 发布日期:2012-02-20
  • 作者简介:王风娥(1981－)，女，硕士研究生，主研方向：中文信息处理；谭红叶、钱揖丽，副教授、博士
  • 基金项目:
    国家自然科学基金资助项目(61100138, 61005053)；山西省高校科技开发基金资助项目(20091001)；山西省自然科学基金资 助项目(2011011016-2)

Recognition of Temporal Relation in One Sentence Based on Maximum Entropy

WANG Feng-e, TAN Hong-ye, QIAN Yi-li   

  1. (School of Computer and Information Technology, Shanxi University, Taiyuan 030006, China)
  • Received:2011-08-17 Online:2012-02-20 Published:2012-02-20

PDF

1307

被引次数

21 | 5

摘要： 分别对句内事件-时间对关系以及事件对之间的时间关系识别进行研究。分析影响时间关系识别的语言特征，如时间关系对之间的依存关系序列、间隔词数、信号词及其位置等，并使用基于最大熵的方法进行识别。实验结果表明，运用该方法获得的事件-时间对关系识别准确率为87.83%，事件对之间的时间关系识别准确率为80.79%。

关键词: 时间关系, 句内时间关系, 最大熵, 依存分析, 自然语言处理

Abstract: This article studies the recognition of the temporal relation between an event and a time expression and the temporal relation between two events where one event syntactically dominates the other event in one sentence. It analyzes some effective linguistic features such as dependency parsing information, relative positions, signal words, position of signal words and so on. A method based on maximum entropy model is proposed. In addition, how linguistic features could affect temporal relation recognition is analyzed. Experimental results show the accuracies of the two tasks are respectively 87.83% and 80.79%.

Key words: temporal relation, temporal relation in one sentence, maximum entropy, dependency analysis, natural language processing

中图分类号: