| Network public opinion is an aggregation of the public’s opinions,sentiments and attitudes for a social event,based on the internet platforms.In the mobile environment,with the development of new media platforms like Sina Weibo,the public tend to express their opinions through the Internet,making the Internet platform the main medium of public opinion.On the one hand,online public opinion has a positive role in promoting hot social events and can accelerate the development of society;on the other hand,negative online public opinion can easily lead to mass incidents,resulting in thorny problems for the supervision of related departments.Compared with the traditional public opinion management and control process,Weibo public opinion control in the era of big data is much more difficult.Therefore,it is necessary to build a comprehensive network public opinion monitoring system for the Weibo platform based on data mining methods and to provide targeted advice for the related departments by monitoring the development of online public opinion.Based on the text mining method,this article takes “5G commercial use” Weibo data as an example to analyzes the sentiment tendency and evolution analysis of the network public opinion of this hot event.This article mainly includes: Firstly,using of Python web crawlers to collect hot Weibo topics and comment data containing "5G commercial use" content to analyze the global situation of public opinion from the hot analysis and feature extraction of online public opinion;Secondly,based on the enrichment and improvement of multiple mainstream emotional vocabulary ontology libraries,an emotional dictionary for Internet public opinion can be constructed.At the same time,based on the Word2 vec algorithm,the extended emotional vocabulary can be marked with sentiment scores.Combining with Negative words,sentence patterns,Weibo emoticons and other factors,an emotional tendency analysis model can be built,to present the sentiment word frequency,sentiment type,and sentiment polarity intensity of Weibo topics and comments through text visualization.Thirdly,the evolutionary process of “5G commercial use” network public opinion is explored through cluster analysis,and the public opinion characteristics of Weibo topics at various stages are summarized,providing targeted reference suggestions for the subsequent guidance of the emotional evolution of network public opinion.This empirical results show that the topic of "5G commercial use" Weibo is relatively high.Its public opinion covers most parts of the country.Generally,it shows positive emotional tendencies,but most of them are mildly emotional.Cluster analysis will help Weibo topics are grouped into six categories,mainly reflecting 5G’s commercial products,user experience,industry development,life impact,patented technology,application prospects and other related content.From the generation to the extinction of each type of Weibo topic in this event,it can be roughly divided into three phases: beginning,erupting,and receding.Among them,the media and netizens’ attitudes towards hot events at the beginning stage are relatively complicated.Positive public opinion disclosure can play a good role in promoting this stage,and the subsequent development of online public opinion can be predicted macroscopically through emotional tendency analysis and opinion extraction;The duration of the outbreak phase is short but has the highest degree of influence.The public opinion characteristics of various topics during the outbreak phase can basically reflect the general public opinion characteristics of the social event.Related departments should pay more attention to monitoring and guiding online public opinion during the outbreak phase.The influence of microblog topics and user participation are usually low in the dissipating stage,but hot events should still be tracked and reported,while microblog rumors and false comments should be avoided,a good microblog communication environment can be established,and the second outbreak of negative online public opinion events. |
