Research On The Evolution Of Network Public Opinion Emotion Of Public Events Considering The Views Of Opinion Leaders

With the rapid development of Internet technology,people are used to participating in the discussion of public events through social networking platforms in daily life.The continuous spread and fermentation of public opinion on the Internet can directly or indirectly impact people’s views and emotions on a large scale,and may even affect the stability and security of the country and society.Therefore,this study has important practical significance to mine the public opinion emotion of public events and analyze the emotional evolution,so as to provide reference for the government public opinion supervision department to monitor and manage public events online public opinion scientifically and efficiently.The specific work is as follows:From the perspective of considering the views of opinion leaders,this study constructs the analysis framework of public opinion emotion evolution of public events network,which is divided into five parts: data collection and preprocessing,public opinion life cycle division,opinion leaders’ view mining,theme mining and emotion analysis.First,the data collection and preprocessing module includes crawling the data relate d to specific public events of Sina Weibo,and preprocessing the experimental data with Python programming language.Second,the division of public opinion life cycle is to divide the stages of event public opinion life cycle according to the distribution characteristics of event data in time series according to the theory of public opinion life cycle.Thirdly,the opinion mining module of opinion leaders includes the construction of opinion leader identification model based on entropy weight method and stance detection model based on Bert model.Identify a set of representative opinion leaders,and screen the content of efficient use of public opinion from the views of opinion leaders based on the detection results of comment stance information.Fourth,the topic mining module is to build a topic mining model integrating sentence features to mine core topics from efficient public opinion content.Fifthly,The emotion analysis module includes building a theme level emotion model,mining the emotion intensity of the same theme,analyzing the theme level emotion of public events in different stages of the public opinion life cycle,and studying the characteristics and laws of theme level emotion changes.Taking the " The incident of a girl jumping off the truck of Company Lalamove " as an example,the data mining results show that the opinion leader set found in t his study is representative,the extracted topics largely represent the essence of the subject,and the mined topic level emotions can better represent the evolution trend of the topic and emotion.Generally speaking,the theme level emotion experienced si x stages from outbreak to calmness;There were two outbreaks of public opinion heat in the whole process;The emotion under all themes is dominated by negative emotion,and the change of theme and the fluctuation of emotion show a certain regularity.This study analyzes the evolution of network public opinion of public events by integrating a variety of algorithms.Theoretically,it complements and perfects the theories and methods of network public opinion governance,and also provides new research perspectives and ideas;In practice,it has a certain reference significance for the public opinion governance decision-making of the government public opinion supervision department.