Research On The Portrait Of High-impact Users On Social Media In The Context Of Emergency Situation

The rise of opinion leaders on social networks has a profound impact on the public,and high-impact users on social platforms are attracting more and more attention.The microblogs published by high-impact users on these social platforms may lead the public opinion of emergencies,and have positive or negative effects on the later evolution of events and their results.Identifying high-impact users on social platforms is of great practical significance in grasping the network public opinion of emergencies and guiding the direction of public opinion.To this end,this paper proposes a method to identify high-impact users of social media based on topic consistency and emotional support.This method takes "He Jiankui Gene Editing Infant Event" as an example.Firstly,Biterm topic model is used to mine the topics of all microblog texts,and each microblog text is combined with their comments.For each microblog-comment combination file,Biterm topic model is used to obtain the topic probability distribution matrix of the new corpus.Comparing the difference of the values in the probability distribution matrix of the topic to determine the consistency between the text of the microblog and the topic of the received comments,and the topic correction can be completed.Then,by using SVM-based emotional classification method,the emotional inclination categories of microblog comments are obtained,the emotional support attitude of the commentators to the original microblog text is determined.After determining the emotional support attitude of all comments received under a microblog text,whether the user who publishes the micro-blog is a positive or negative high-impact user can be identified.Nextly,we construct a group of high-impact user characteristics indices.The high-impact user characteristics indices selected in this study are: user basic information indices,including account level,account authentication,number of followers,number of fans,number of microblogs,personal profiles,and so on;basic attribute indices of microblogs,including the number of being liked,number of comments received,number of being forwarded,average length of comments,support degree,and so on.Then the weight is given to some indices to calculate the value of user influence,and we determine the opinion leaders with positive and negative high influence according to their values of user influence.On the basis of opinion leaders,this paper introduces the factor of "microblog publishing time",which divides the development and evolution of the event public opinion into different time periods.We explores the composition of opinion leaders in each time period,and finds that in the later period of the event,some opinions are obviously inconsistent with those in the earlier period of "mainstream" opinions,that is,there are some opinion reversers.Finally,a complete portrait of Weibo users is constructed by combining static information data of Weibo users' homepage with dynamic behavioral characteristic data.This research can provide management departments with netizens' focus of attention and emotional attitude to the public emergencies,help them grasp the online public opinion in time,and provide effective support for decision-making and management for emergencies.Through the portrait construction of high-impact users on social network of emergencies,this research is helpful for relevant departments use high-impact users to publish some information to guide the public opinion according to the characteristics of these users.This work was supported by the Major Project of the Ministry of Education of China(Grant No.17JZD034).