计算机工程与应用 ›› 2009, Vol. 45 ›› Issue (2): 131-133.DOI: 10.3778/j.issn.1002-8331.2009.02.038

• 数据库、信号与信息处理 • 上一篇    下一篇

中文词语倾向性分析处理

李 娟1,2,张 全2,贾 宁1,2   

  1. 1.中国科学院 研究生院,北京 100039
    2.中国科学院 声学研究所,北京 100190
  • 收稿日期:2008-07-03 修回日期:2008-10-17 出版日期:2009-01-11 发布日期:2009-01-11
  • 通讯作者: 李 娟

Semantic orientation identification for Chinese opinion terms

LI Juan1,2,ZHANG Quan2,JIA Ning1,2   

  1. 1.Graduate University of Chinese Academy of Sciences,Beijing 100039,China
    2.Institute of Acoustics,Chinese Academy of Sciences,Beijing 100190,China
  • Received:2008-07-03 Revised:2008-10-17 Online:2009-01-11 Published:2009-01-11
  • Contact: LI Juan

摘要: 意见挖掘是自然语言处理研究领域的一个新热点。词语倾向性的判定是意见挖掘的基础和重要环节。该文进行了中文词语倾向性的自动判定实验。实验中采用了《现代汉语褒贬用法词典》中的词语做为褒贬判定的核心词汇,以同义词词典扩展了褒贬义词典的词语,并使用二元语法模型来判定多倾向性词语的倾向。实验结果褒义词的F-Score为79.31%,贬义词的F-Score为78.18%。

关键词: 意见挖掘, 词语倾向, 二元语法

Abstract: Opinion mining is a new hotspot in the area of natural language processing.Determining the opinion orientation of the glossary is a foundation and very important component in an opinion mining system.An experiment is carried out on opinion orientation identifying for Chinese opinion terms.In the experiment,the authors take the words which are in COMTEMPORARY CHINESE LANGUAGE ORIENTATION USAGE DICTIONARY as the seed words,and extend them by synonyms dictionary.Further more,Bigram theory is adopted to disambiguate the multi-orientation for one word.The F-score of the experiment reaches 79.31% for positive words and 78.18% for negative words.

Key words: opinion mining, semantic orientation, 2-Gram