| In recent years,many netizens have commented on hot events,reflected people’s livelihood and made suggestions and suggestions through Internet channels.Online social platform has become an important way to promote the construction of socialist democracy.However,sudden disasters,public health and safety incidents,etc.,have caused serious threats to people’s lives and huge losses to property safety.After the incident,a large number of real and false information such as relevant real news and fabricated rumors were mixed,and then fermented through social network platform media,causing a huge social panic and seriously threatening social security.Therefore,in combination with the current situation of public opinion management on social networks,we should promote the combination of statistical and machine learning methods,correctly analyze user behavior characteristics and relationships,identify the emotional categories of public opinion text data of microblog in emergencies,timely understand the emotional tendencies of Internet users,and judge the impact of microblog users in hot events.It plays an important role in guiding the development of Internet public opinion and maintaining social stability.In this paper,python will use the open API platform of sina Weibo to crawl and process the popular Weibo data and comment data related to the "Changchun Changsheng problem vaccine" event spread on the Weibo platform,including: Weibo user ID,its Weibo content,comment content.Due to the limitation of sina Weibo API,the octopus data collector is also used to supplement user information data,including:related user information of Weibo bloggers(number of followers,number of fans,number of published Weibo),popular Weibo forwarders,commentators and related user information.Next,the data is stored,de duplicated,cleaned and other pre-processing work,and the distribution of groups and hot spots of public opinion is counted to judge the trend of real-time public opinion.Starting from the further collection of user relationship characteristics such as the number of microblogs,comments,and forwarding numbers,and using SPSS software to analyze the data,it is found that microblog users’ behavior is more influential than praise and comment forwarding.Through the example analysis of this paper,there is a negative correlation between the number of microblog users’ attention and the number of fans,which is not absolutely correct with the previous research.It shows that the more attention microblog users pay,the more disorder they have,the less fans they have.For the sample text data processing,we extract the blog content and comment content from the data set for Chinese word segmentation,stop word filtering,feature processing and other text analysis,and then use the naive Bayesian classifier in machine learning method for emotional analysis.The P,R and F values of naive Bayes classifier are 84.04%,79.91% and 81.93%,respectively.It can be concluded that this classification method is ideal.Finally,using analytic hierarchy process and Delphi method to construct the real-time influence evaluation system of microblog public opinion events in four aspects of microblog activity,attention,interaction and emotional guidance,calculate the weight of each factor,and comprehensively get the real-time influence of users in the process of microblog public opinion communication.It is found that the influence of negative emotion is greater than positive emotion,and the greater the polarity of emotion,the greater the influence of micro blog communication.Combined with the development trend of microblog public opinion,in order to better control the dissemination of information related to public opinion events,facilitate the understanding of the influence of true and false information and the true feelings of users,we can put forward research suggestions with reference value to social network public opinion early warning to a certain extent. |
PDF Full Text Request