一种基于知网的中文句子情感倾向判别方法*

针对基于知网的中文句子情感倾向判别方法中存在的准确率不高的问题，提出采用否定模式匹配与依存句法分析相结合的方法。研究分析了修饰词极性以及否定共享模式，确定修饰词以及扩展极性的定量和否定共享范围，提出依存语法距离的影响因素来计算情感倾向，并且在否定模式匹配后改进句子极性算法。实验结果表明该方法取得了良好的效果。

Method of discriminant for Chinese sentence sentiment orientation based on HowNet

This paper proposed a method which combining negation model matching with dependency parsing , to solve the low precision rates in the Chinese sentences sentiment orientation based on HowNet. On one hand, according to analysis the negative polarity of the qualifier and sharing models to determine the polarity of the qualifier, identified the expansion of quantitative and the scope of negation sharing. On the other hand, according to the dependency grammar distance factors to calculate the sentiment orientation, and to improve the polar algorithms of the sentence after the negation model-matching. The experimental results show that the method has achieved good results.