Analysis Of Weibo Public Opinion On The "MU5735" Incident Based On Text Conten

On March 21,2022,a sudden social and public incident that touched people across the country occurred-the accident of China Eastern Airlines MU5735 in Wuzhou,Wuzhou County,Guangxi.Countless people followed the latest development of this incident,and also posted their feelings and exchanged their opinions on major online social media platforms.Weibo is one such social media platform and press release platform that carries the voices of countless netizens.In the face of all kinds of speeches and the huge amount of information in the text of comments,it is especially important to use suitable analysis methods to sense people’s emotional tendencies,so as to monitor and guide public opinion.This paper uses web crawler technology as a tool and the microblogging platform as a data source to crawl the content of blog posts with the keyword ”MU5735”published by the official microblog of the People’s Daily and the corresponding comments,and conducts a series of analyses and draws certain conclusions and recommendations on the public opinion situation and public opinion guidance.The research is divided into three parts,namely data collection,pre-processing and Chinese word separation,sentiment analysis and descriptive analysis in stages,and LDA topic modelling.In the data collection,pre-processing and Chinese word separation,this paper mainly uses the python request library to obtain the original microblogs and comments from microblogging platforms,and then carries out data pre-processing work such as data cleaning,custom dictionary Chinese word separation and deactivation of the obtained microblogs and comments to obtain more standardised text data.In the sentiment analysis and phased descriptive analysis,this paper used the snow NLP library to analyze the data for sentiment,and divided the opinion development into three phases based on sentiment scoring,namely from March 21 to March 26 at 21:00,from March 26 at 22:00 to before April 20,and after April 20.Afterwards,descriptive statistical analyses such as drawing word cloud diagrams were conducted on the content of blog posts and comments in stages according to the three divided stages.In the LDA topic modelling,this paper focuses on the comment content to carry out the phased LDA topic modelling.The study shows that in the first stage,netizens’ emotions fluctuate more and are more negative.At this stage,the official media released the latest news in a timely manner and took the initiative to popularise relevant professional knowledge,which helped to pacify the sentiment of the netizens.In the second stage,thanks to the official media’s efforts,netizens accepted the search and rescue result that all the people were killed well and looked forward to the early recovery of the truth about the accident.In the third stage,netizens were more dissatisfied with the content of the report,which did not reveal much new information,and public opinion was most negative.Finally,the study is summarised and some recommendations are made on how to manage public opinion in the event of similar incidents in the future.