融合多种情感资源的微博情感分类研究

为了通过融合多种情感资源库中的词汇情感特征来提高微博情感分类精度,提出了一种词汇情感确定性度量的计算方法,并以此为基础将在多种情感词汇上获取的情感特征融合为词汇的综合情感特征,然后采用机器学习的分类方法实现微博观点句识别和观点句情感倾向性判定.实验表明,本方法利用词汇的情感确定性度量,统一了词汇情感倾向性的强度度量,在观点句识别和观点句情感倾向性判定两个情感分类任务中都取得了较好的性能.

Research of Microblog Sentiment Classification Based on Emotional Polarity Certainty for Vocabulary

To improve the sentiment classification precision,the paper proposed a method to determine the certainty of sentiment.The method takes account of multiple sentiment features of vocabularies in the sentiment database and identifies the clause giving opinions.It classifies the clauses into different emotinoal polarities through machine learning.The experiment shows that the proposed method employing the certainty of the sentiment unifies the emotional polarities measurement.It has a better performance on identifying the clause giving options and emotional polarity classification.