摘要
情感特征抽取是文本情感分类的重要步骤,正确的选择情感特征并赋予合理的情感权重是保障分类精度的前提。利用基础情感词词典、连词词典及词语距离,提出了一种基于多重词典的中文文本情感特征抽取算法,实验证明该方法优于HM,SO-PMI和词语语义距离等经典的特征抽取算法。
Emotional feature extraction is an important step in text sentiment classification,so choosing emotional feature correctly and giving a reasonable sentiment weight are the premise to guarantee classification precision.A Chinese text emotional feature extraction algorithm is proposed based on multiple lexicons including basic semantic lexicon,conjunction lexicon and word distance.The experiment results show that the algorithm outperforms some classic feature extraction algorithms of HM,SO-PMI and word semantic distance etc.
出处
《湖南工业大学学报》
2011年第2期42-46,共5页
Journal of Hunan University of Technology
基金
湖南省自然科学基金资助项目(10JJ3002)
中国包装总公司科研基金资助项目(2008-XK13)
关键词
情感特征
情感权重
多重词典
情感特征抽取算法
emotional feature
sentiment weight
multiple lexicons
emotional feature extraction algorithm