Research On Bursty Event Detection And Traceability Analysis In Social Media

In recent years,social networks have been continuously popularized,and various social media have emerged in an endless stream.As one of the most interactive and real-time mainstream social media,Weibo provides users with a platform for free speech and access to other sources of information,thus forming Massive data.Many bursty events in the real world are published on Weibo,and there are no perfect supervision measures on Weibo.During the dissemination of bursty events,false information such as false information and rumors will be formed.Such information will be polluted.The network environment,through continuous spreading,will amplify the influence of false speech and form a situation.Amplifies the negative impact of false speech on society.Therefore,carrying out bursty events detection and traceability research on Weibo can keep abreast of the dynamics of social public opinion,and at the same time can grasp the source of the event,helping the government and other relevant departments to control the spread of the event from the source.Because Weibo contains irregular text,the structure is not uniform,and large amount of data,in view of the emergent nature of bursty event,this article proposes a hot topic-based Weibo bursty event detection method.For the suddenness of the event,the value of the sudden word is calculated based on the data of the two days before and after,the word with high sudden value is selected as the key sudden word,and a co-occurrence matrix is ??established for these key sudden words.The current matrix can calculate the dissimilarity matrix of the key burst words,and then use the agglomerative hierarchical clustering algorithm to get the hot topics,use the discrete naive Bayes classification algorithm to train a classification model,and use this model to perform the hot topics Classification to detect bursty events and their types.Aiming at the problem that the source of bursty event is not necessarily the user with the earliest release time,but that the more influential Weibo becomes the source of the event,the more likely it is that Weibo is the source of the incident.This article proposes a Weibo emergent incident traceability based on influence analysis Methods.This method analyzes the influence of Weibo users and the influence of Weibo events.User influence is analyzed from the perspectives of opinion leaders,user fans,user ratings,and user authentication weights.Event influence is analyzed from Weibo.Analyze the content's release time,originality,text similarity,number of reposts,number of comments,etc.,and finally find the source of the incident based on the influence of these two aspects.Based on the analysis of the experimental results,the proposed method can effectively filter general hot topics,detect bursty events and their types;the proposed micro blog emergency Traceability Method Based on impact analysis has a high accuracy for the source discovery of incidents.