A new quantitative sentiment analysis approach based on sentimental dictionaries and machine learning is proposed in this paper to tackle the problem of partiality of quantitative sentimental features of representation for objective facts.The positive and negative emotional weights and probabilities obtained by the combination of emotional dictionary and machine learning are used as new emotional characteristics.These emotional characteristics are combined with the language,attributes and information characteristics of the commentary text to redefine the affective characteristics of the user's behaviour.Experimental results show that the support vector machine (SVM) algorithm can classify the subjective and objective based on emotion characteristics,and the accuracy can reach 87.20%.%针对描述客观事实评论中量化的情感特征片面问题,提出一种基于情感特征的主客观分类方法.将基于情感词典与机器学习结合得到的积极或消极情感权值与概率,作为新的情感特征项与评论文本的语言、属性和信息特征相结合,重新确定影响用户行为的情感特征,从而对评论文本进行主客观分类.实验结果表明,采用支持向量机算法可使基于情感特征的主客观分类效果更佳,准确率为87.20%.
展开▼