图书情报工作 ›› 2010, Vol. 54 ›› Issue (11): 104-127.

• 信息技术 • 上一篇    下一篇

句子情感分析及其关键问题

李纲,程洋洋,寇广增   

  1. 武汉大学信息管理学院
  • 收稿日期:2010-01-13 修回日期:2010-05-06 出版日期:2010-06-05 发布日期:2010-06-05
  • 通讯作者: 寇广增
  • 基金资助:

    文本集特征提取方法及应用研究

Key Problems of Sentence Level Sentiment Analysis

Li Gang 1,Cheng Yangyang 2,Kou Guangzeng 1   

  1. 1. School of Information Management of Wuhan University,
    2. Information Management School of WhuHan University,
  • Received:2010-01-13 Revised:2010-05-06 Online:2010-06-05 Published:2010-06-05
  • Contact: Kou Guangzeng

摘要:

情感分析关注具有情感倾向的评价性信息,具有广泛的应用。情感分析按照粒度的不同分为三种:词汇情感分析、句子情感分析和文档情感分析。文中对句子情感分析及其关键问题进行介绍,首先简要描述句子情感分析的任务,然后介绍句子情感分析中主客观句分类方法及两种主观句情感分类方法——基于情感词的方法和机器学习方法,最后对情感分析中的三个关键问题——词汇上下文极性判定、评价主题识别、意见持有者识别进行总结。

关键词: 句子情感分析, 词汇上下文极性, 评价主题, 意见持有者

Abstract:

Sentiment analysis studies on evaluative information, it has a wide range of application. Divided by granularity, there are three kinds of sentiment analysis, including word, sentence and document level sentiment analysis. This paper mainly focuses on the sentence level sentiment analysis. First describes its tasks briefly, then introduces the methods of sentence S/O classification and two ways of subjective sentence sentiment classification, finally summarizes three key problems of sentence level sentiment analysis: word contextual polarity disambiguation, topic identification and opinion holder identification.

Key words: sentence level sentiment analysis, word contextual polarity, evaluation topic, opinion holder