北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2012, Vol. 35 ›› Issue (5): 90-93.doi: 10.13190/jbupt.201205.90.suny

• 研究报告 • 上一篇    下一篇

基于依存关联分析的情感词扩展

孙 艳, 周学广, 付 伟   

  1. 海军工程大学 电子工程学院
  • 收稿日期:2012-01-10 修回日期:2012-05-21 出版日期:2012-10-28 发布日期:2012-07-06
  • 通讯作者: 孙艳 E-mail:yzisun@126.com
  • 作者简介:孙 艳(1983-),女,博士生,E-mail:yzisun@126.com 周学广(1966-),男,教授,博士生导师
  • 基金资助:

    国家自然科学基金项目(611100042)

Opinion Words Expansion Based on Dependency Parsing

SUN Yan, ZHOU Xue-guang, FU Wei   

  1. College of Electronic Engineering, Navy University of Engineering
  • Received:2012-01-10 Revised:2012-05-21 Online:2012-10-28 Published:2012-07-06
  • Contact: Yan SUN E-mail:yzisun@126.com

摘要:

无扩展的情感词识别方法对于倾向容易发生演化的社区媒体信息的召回率很低.基于此,提出了一种基于依存关联分析的情感词扩展识别方法,首先对文本进行新词和短语发现;再对句子的依存树进行剪枝和归并;然后按照给定的策略进行情感词和评价对象的扩展;最后得到情感词扩展的识别结果.该方法能同时扩展具有直接依存关系和间接依存关系的情感词. 实验结果表明该方法有效.

关键词: 依存分析, 情感词扩展, 倾向分析, 舆情

Abstract:

The forgoing experiments have shown that the method of opinion words identification with non-propagation is extremely limited on social medium information. A method of opinion words expansion identification based on dependency parsing is presented. Through new words and phrases discovery, pruning and merger on dependency tree, the opinion words are expanded. A key advantage of the proposed method is that it can expand words in direct dependency and indirect dependency. Experiments verify the feasibility of this method.

Key words: dependency parsing, opinion words expansion, polarity analysis, public opinion

中图分类号: