| With the advent of the Internet era,the government and society become more and more attention to the "network public opinion".Users can freely make comments on the Internet with their opinions,attitudes,and emotions,even express their subjective affection views.Activities on Micro-blog such as posting,commenting,and sharing always reveal their personal views and relevant feelings.With the continuous fermentation of network public opinion,personal opinions will become the"barometer" of network public opinion,affecting the dissemination of network public opinion.This paper adopted the event of "Northward Migration of Yunnan Elephant Treks",which hotly debated topic on Sina Micro-Blog in 2021,based on the online public opinion data on the Blog.Meanwhile,we can reveal the characteristics and the influencing factors of public opinion dissemination on the website media platform,to provide suggestions for the relevant departments to take appropriate measures to lead the healthy direction of the online public sentiments.The main research work includes:Firstly,we collect 44501 text data and perform data preprocessing such as Chinese word segmentation,part-of-speech tagging,and word weight statistics.Secondly,this experiment uses two supervised machine learning algorithms of Naive Bayes and SVM for emotion classification of the Chinese text.The results showed Naive Bayes slightly outperformed SVM.Thirdly,we divide the period of public opinion evolution into the initial stage,the growth stage,the outbreak stage,and the recession stage.Then we can explore hot topics by TF-IDF algorithm and use comparison charts to study the emotional tendency of public opinion in each period.Finally,the regression analysis is used to explore the influencing factors of public opinion dissemination,such as subjective emotion,information timeliness,popularity,users’ preference,influence,and activity of information publishers.This paper provides new methods and ideas of sentiment analysis,text mining,and digital visualization.That has some actual value.The sentiment analysis method was selected,providing a reference for identifying subjective emotions in Chinese texts.Meanwhile,the relevant conclusions supply empirical support for exploring the characteristics and the influencing factors of public opinion dissemination and enrich the existing research results. |
