Abstract: It is significant to analyze emotional uncertain words and sentences in Chinese text sentiment analysis.Emotional uncertain words are generally words with rich meaning,which implies some evaluation in the expression.Emotional uncertain sentences usually have the same size of positive words and negative words,so the emotion is not ob-vious.In this paper,using uncertain word “好” as an example,we designed features for the uncertain sentences.Then using four different classification algorithms of supervised learning to do experiments,we got the conclusion:SVM can better deal with the emotional uncertain words and sentences.

Key words: Sentimental analysis,Uncertainty,Supervised learning,SVM