针对基于词典的传统分类器无法对不在词典中的情感词的极性和强度进行有效计算和细分的问题,基于最大期望模型,提出构建完善情感词典的EM-SO算法,在此基础上设计基于语义倾向计算模型的否定式和强(弱)化处理组件,以获取评价词及其修饰词的组合效应,实验结果表明,所提算法及所设计组件在评论集上对情感词极性和强度的计算性能优于SO-CAL模型,可应用到主观性分类等实际任务中.%For the polarity and strength of unknown sentiment words cannot be calculated and classified effectively in traditional lexicon-based classifiers. The EM-SO algorithm was proposed based on expectation maximization for constructing and updating sentiment lexicon. Negative and intensifying components were designed upon SO-CAL for capturing the combined effects of appraisal words and their modifiers. Experimental results showed that the EM-SO algorithm and designed components outperform SO-CAL obviously for the calculation performance of the polarity and strength of sentiment words on review sets and can be applied to the subjective classification task and so on.
展开▼