运用非结构化信息挖掘,对网络评论情感进行分析是一个非常重要的方法。本文基于Web客户评论情感文本,在情感文本预处理过程中使用四种不同的停用词表,采用两种不同的特征选择方法,选用著名的TF-IDF权重计算方法,使用基于RBF核函数的支持向量机方法的分类器实现了对携程网上采集的4000个酒店客户评论情感文本的分类研究。通过实验,分析了不同特征选择方和停用词表的使用对客户评论文本情感分类的影响,提出了基于情感文本分类的有效的停用词表。
It is an important method to analyse Web reviews' sentiment categorization with unstructured information date mining.This paper based on the Web text reviews,using four different kinds of stop word removal way,two kinds of feature selection methods,the famous TF-IDF weighing assignment methods and the SVM(support vector machine) technology with the RBF kernel function categorize the 4,000 customer reviews text grasp on XIECHENG.With the results of the experiment,this paper analysis the influence of different kinds of feature selection methods and stop word removal on the Chinese text sentiment classification,represent the more effective stop word removal list.