网络环境下中文情感倾向的分类方法

Categorization of Chinese Emotional Reaction Types on the Net

论文从中英文语言差异的角度出发,针对语义倾向方法在中文应用中暴露出来的问题,提出了具体的应用于中文的改进算法。在实验中,运用基金测试文档进行实测以验证改进算法的有效性。实验表明,改进后语义倾向方法在应用于网络环境下中文文本情感倾向分类中具有理想的性能,并具有不需要大量训练样本、对领域知识有较弱的依赖性等特点,展示出良好的应用前景。

In this paper, we propose a Chinese emotional reaction categorization method on the Net based on semantic preference. An data mining technique which is based on semantic preference is used. In terms of difference between Chinese and English, we analyze the problem of semantic preference when applied in Chinese and propose an improved algorithm. In the experiments, new algorithm is applied to Chinese fund reviews. The results show that improved algorithm performs better when it is used to Chinese emotional reaction categorization method on the Web. Compared with the trained classification algorithm, the improved algorithm doesn＇t need to be trained with a great deal of documents and has little dependence on domain knowledge, which promises a bright future application.