Abstract:

As the rapid spread of the internet, people have gotten used to communicating with others through internet. However, because of the rapid exchange of information and the lack of social feedback and social norms, people become more free and extreme, and the expression of emotion is more real, which leads that people intend to focus on information itself and ignore the social regulations. Remarks written by net citizens are always negative because the emotions they try to express are negative. Moreover, the accumulation of negative emotions on internet will develop into social crisis in reality. This article primarily introduces the emotional analysis of Web textual data by using Web text mining technology and EM algorithm based on native Bayes classifier, which divides emotions on internet into positive emotions, neutral emotions and negative emotions. Meanwhile, the emotional analysis method is able to take precautions against social crisis in reality by the classification, analysis and early warning of negative information.

Key words: Web text mining, sentiment analysis, real hazard