Public Opinion Analysis Of Hot Events Based On Text Mining

A sudden epidemic in 2020 disrupts everyone's rhythm.In order to alleviate the impact of the epidemic on the society,ensure the normal operation of the economic market,stimulate consumption,and alleviate the employment situation,the stall economy returns to the public view again.Since May 27,2020,when the land stall policy was relaxed,the topic of stall economy has reached its peak.Many people have responded positively to the call of the state and opened their own way to set up stalls.The stall economy has become a hot topic among the public.However,in less than half a month,the economy of the stall has cooled rapidly again,and it has returned to its original state.In less than a month's time,the stall economy has developed from full swing to rapid cooling.Analyzing the public opinion topics about stall economy in this period will help us find the problems of stall economy and develop stall economy more reasonably.This paper takes "stall" as the key word,uses web crawler technology to crawl the user micro blog data from May 26,2020 to June 30,2020,a total of 22368,and carries out text mining analysis on this data.By using word cloud image and semantic network analysis,the focus of public attention in each stage is mined,which lays the foundation for the follow-up study of people's emotional tendency.Four classification models are used for modeling analysis,and relevant indicators are used to evaluate the advantages and disadvantages of the model.Finally,the Ada Boost optimal model is selected to predict the trend of public sentiment.Most of the public hold rational views on the stall economy.Relevant departments can make corresponding adjustments according to the contents of public concern and public feelings,and scientifically guide the development of stall economy.The main contents and conclusions of this paper are as follows:(1)Using "stall" as the keyword of microblog search,using web crawler technology to crawl users' microblog data from May 26,2020 to June 30,2020,a total of 22368 microblog data were crawled.According to the original microblog data crawled,the following research and analysis were carried out.(2)The text data is preprocessed,including data cleaning,removing stop words,text segmentation and so on.Word2 vec method is selected for text vectorization.(3)Semantic mining analysis of text data,statistical analysis of word frequency,making word cloud map of each stage,making semantic network map,sorting out high-frequency Related words,analyzing the content of public concern,finding the relationship between highfrequency words,further mining the emotion of the text,providing the basis for the follow-up emotional research.(4)The emotional content of the text is divided into three categories: negative,neutral and positive.This paper selects Adaboost,GBDT,XGBoost and LSTM algorithms to train the text classification.From the training results,Adaboost model performs best.Therefore,this paper selects Adaboost model to predict the remaining text,and integrates the prediction results to draw a conclusion.In the three stages of development,the proportion of neutral emotion is the most,and the proportion is higher and higher.Most people hold neutral views on the land stall economy,and gradually lose enthusiasm for the stall,and all of us gradually return to rationality.