Research On Evaluation And Trend Forecast Of Weibo Public Opinion Heat Based On Sentiment Analysis

In the era of big data,online public opinion is developing explosively,and the governance of online public opinion has become an important part of the modernization of our country’s governance capacity.Nowadays,Weibo has developed into the main driving medium of online public opinion.Accurately evaluating and timely predicting the trend of fluctuation of Weibo public opinion can help provide theoretical support and reference for the government or enterprises to determine the priority of public opinion response and formulate guidance strategies.Based on the original interactive data such as the number of reposts,comments,likes,and readings that are commonly used on Weibo platforms to evaluate the heat of Weibo public opinion,it is easily interfered by the "pseudo-hot spots" problems caused by the increasingly flooded paid posters and fan size.The amount of interactive behavior data is high,but the topic value is low,and it does not have the value of public opinion control.Its text content is single and the emotional intensity is weak.Based on the previous research,this paper introduces the emotional intensity of public opinion content into the Weibo public opinion heat evaluation system on the basis of the influence of the main body and the influence of dissemination,and constructs a Weibo public opinion heat index system based on sentiment analysis.The sentiment intensity in the dimension of senti mental tendency is quantified by sentimental dictionary-based sentiment analysis method.In order to further accurately measure the emotional intensity,an extended emotional dictionary is constructed through the SO-PMI algorithm,and more fine-grained emotional quantification is carried out for emotional words in different semantic environments of Weibo texts.In order to avoid relying on the subjective experience of experts and the difficulty of monitoring in practice and other problems affecting the eval uation efficiency,the entropy weight-TOPSIS method is selected to objectively evaluate the heat of Weibo public opinion to study and judge the evolution trend of the heat of public opinion,and to realize the Real-time monitoring function.In the past,the BP neural network model was commonly used to predict the trend of public opinion.In this paper,the genetic algorithm and the improved sparrow search algorithms is used to optimize it.This paper selects the "Sudden Death of Pinduoduo Employee" Weibo p ublic opinion event for empirical research,and finds that:(1)The results of the entropy weight-TOPSIS empirical evaluation are in line with the evolution characteristics of the online public opinion life cycle.It can effectively evaluate the heat of Weibo public opinion events.(2)Compared with the evaluation results of the quantitative characteristic indicators of the original interactive behavior,the index system based on sentiment analysis has a certain early warning ability in evaluating the heat of Weibo public opinion during the outbreak period,and it can better reflect the actual changes in facts,and the evaluation accuracy is obvious advantage.(3)The BP neural network model optimized by the improved sparrow search algorithm is better than t he genetic algorithm,and can more accurately predict the trend of Weibo public opinion.(4)Compared with the prediction result of the evaluation value of the original interactive behavior quantitative characteristic index,the Weibo public opinion heat prediction effect based on sentiment analysis evaluation is better.This paper has important theoretical and practical significance for public opinion evaluation,improves the accuracy of Weibo public opinion trend prediction,and provides decision support for Weibo public opinion management and control.