Analysis And Detection Of Rumors Related To New Coronary Pneumonia

The arrival of the 5G era has brought people a faster information interaction experience.The new crown pneumonia epidemic（COVID-19）broke out in China,and there were endless information about the epidemic on social platforms,including some false rumors.Rumors are essentially a virus of public opinion.In the fight against the COVID-19,it is more worthy of vigilance than the real virus.When rumors spread widely,they will cause bad social impact and the harm may be even greater.The generation of rumors reflects the blind spots of information disclosure.This paper conducts data analysis firstly and text mining on rumors of COVID-19 to find out the key content,help relevant departments to share information,and publish information more timely and accurately,so that the rumors lose the opportunity to spread.Regarding the detection of rumors,this paper builds a new and more effective rumor detection model based on the research results obtained by scholars at home and abroad,and applies it to the identification of rumors about the COVID-19.Social networks have become indispensable communication tools in people’s modern life,affecting all aspects of people’s daily life,but the information obtained is not necessarily true and reliable.In response to the COVID-19 pandemic,only real information is valuable to the public and authorities.Therefore,it is very important to detect rumors of COVID-19 on social networks.This paper collects four aspects of data on social platforms as the data set for this experiment: data on Tencent’s true platform,topic data of Weibo "first line of fighting against COVID-19",data of Weibo community management center,and general microblogs related to the COVID-19.We analyze users’ personal information and content information of Weibo related to the COVID-19,extract eighteen features in in four categories,i.e.,text characteristics,user-related features,interaction-based features,and emotion-based features,to build a new rumor detection model.First of all,we use data on Tencent’s true platform and topic data of Weibo "first line of fighting against COVID-19" for text mining to analyze the similarities and differences between the two platforms.Second,by comparing the rumors of the community management center during the epidemic with the rumors of the same period last year,we can explore the characteristics of rumors related to COVID-19.Subject terms are analyzed on the collected Weibo data set to determine the most concerned topics during the epidemic,and topic characteristics of text information are provided for rumor detection.Third,we extract eighteen features of Weibo community management center data and common Weibo data related to the COVID-19,and use Support Vector Machines,naive Bayes,integrated learning and deep learning algorithms to build new models to detect rumors related to the COVID-19 on Weibo.Experimental results show that the four classification models based on eighteen features can reach an accuracy of 82%.Among them,all indicators of the integrated learning model perform well,and the detection accuracy can reach 91%.At the same time,we analyze the importance of 18 characteristic,and the results show that interaction-based features,emotion-based features,and user-related features play important roles in the detection of rumors of COVID-19,which proves the effectiveness of the features selected in this paper.Finally,combined with the research results,we summarize our work and look forward to the future research direction of rumor detection.