一种结合随机游走和粗糙决策的文本分类方法

Text Classification Method Combining Random Walk and Rough Decision

情感分析一直是社交媒体领域所研究的热点,为克服有些情感词语在文本中模糊性强的问题,本文引入了两个模型.随机游走模型在互联网分析及页面排序中有了一些成熟的应用,但在文本倾向性分析中少有涉及.文中提出基于扩展随机游走模型的情感词极性判别算法,对模糊性词语的情感词极性进行分析,通过建立文本向量空间,提出基于情感词极性权重序的属性离散化算法,对候选属性进行离散化处理.最后通过粗糙决策置信度模型,对文本最终情感类别进行判定.实验通过词极性判别、离散化、粗糙决策置信分类三个阶段,把各阶段得到的结果与其他方法进行对比,最后通过多种评价指标对情感分类的最终分类结果进行评判,实验结果证明了方法的有效性.

Emotional analysis has always been a hot topic in the field of social media. In order to overcome the ambiguity of some words in the text,two models are introduced in this paper. Random walk model has been applied more widely in Internet analysis and page ranking,but it is seldom involved in text orientation analysis. In this paper,a word tendency mining algorithm based on extended random walk model is proposed,and the polarity of affective words in fuzzy words is analyzed. By establishing text vector space,an attribute discretization algorithm based on the polarity order of emotional words is proposed to discretize the candidate attributes.Finally,a rough decision confidence model is used to determine the final sentiment category of the text. In the experiment,the validity of the method is evaluated through the three stages of word polarity discrimination,discretization,and confidence classification level of rough decision,and the final classification results are evaluated through a variety of evaluation indexes. The results of the experiment prove the effectiveness of the method.