一种基于粒运算的文本情感分类方法研究

Research of Chinese text sentiment classification based on granule computing

随着Internet的迅猛发展,人们对事件的立场、观点和看法的文本信息每天都会在网上出现,对于这些评论,仅靠人工进行跟踪和分析显然是行不通的,人们开始关注并研究评论文本的主观性情感倾向分析。文本情感分类中,分类器的设计是其中最重要的一个环节。文本评论往往是针对某一个特定领域的产品,评论语句一般都是短短几句,并且词汇量小特征词的交叉比较多,在这种情况下,与那些基于统计方法的分类器比较,基于规则的分类器更具优越性。提出了一种基于粒运算的方法,通过建立粒网络生成分类规则,从而得到评论文本的情感倾向分类。

With the rapid development of Internet,text information containing standpoints,comments and opinions appears on Internet daily.Obviously,it is impossible to track and analyze these comments by hand.People start to pay attention to the subjective tendency analysis of the text comments and begin to do some research on it.In the sentiment classification,design of the classifier is the most important joint.Text comments usually aim at products in a field.The commenting sentences are usually short and brief.The feature words of the comments cross heavily.Under this situation,the classification algorithm based on rule learning has more advantages than the ones based on statistical method.This paper proposes a method based on granularity computing to get the sentiment classification of text comments through rule learning by establishing the gran- ule network