Research on netizens’ emotion evolution in emergency based on machine learning

Abstract

[Purpose/significance] By analyzing the emotional evolution of Weibo users after emergencies, we can find out the law and potential risks of public opinion evolution and provide instruction for the government to control and guide network public opinion. [Method/process] We proposed an analysis model of public opinion evolution based on Emotional Analysis and GBRT. With the help of Python, a web crawler was developed to collect Weibo comments. After that, a Naive Bayesian Classifier was used for emotional analysis. According to public emotion and the number of comments, we divided the evolution process into fever period, persistence period, incubation period and extinction period. Statistical and visualization methods were used to study the evolution characteristics of word cloud, emotional tendency and age groups. Finally, correlation analysis and GBRT were used to predict each individual’s emotions. [Results/conclusion] Taking the dangerous chemical explosion accident in Tianjin as an example, we can validate our model. Results shows that the model can reasonably divide the evolutionary stages, find out the law of public opinion evolution in different stages, and accurately predict users’ emotional tendency.