Research On Rumor Source Estimation Of Online Social Network Based On Community Division

With the rapid advancement of the "Internet +" era,online social networks have been booming in line with the trend of the times.The emergence of online social networks has changed people's social systems and brought people rich and convenient services.At the same time,social networks are full of complicated information,leading to the rapid spread of rumors and other abnormal information in social networks.Rumors not only mislead people's perception,but also affect the stable development of society and the country,causing huge losses.Therefore,how to quickly and accurately identify the source of rumors has very important practical significance.This paper studies the related problems of rumor tracing through complex network theory and rumor spreading rules.The main contents of this paper are as follows:1)Based on the SI propagation model,combined with the network community structure,a dual source tracing algorithm based on spectrum optimization is proposed.Based on the modularity,the algorithm uses an optimized spectral analysis method to divide the infection graph into two non-overlapping communities,and conducts single source source tracing based on rumor centrality in these two communities,and then double sources The source tracing problem is approximately decomposed into two independent single source tracing problems.Finally,by comparing different network structures and different centrality estimators,simulation experiments show that the algorithm's comprehensive traceability performance is better,and the average error distance is within 2.5 hops.Compared with the traditional dual-source algorithm,the algorithm has lower time complexity.2)The problem of multi-source tracing of weighted networks is studied.Due to the weighted network structure of real social networks,the propagation probability is positively related to the intimacy of interpersonal relationships.Therefore,based on the propagation probability,the effective distance is used to quantify the strength of the interaction between nodes,that is,the weight of the network,thereby reflecting the true order of nodes being infected.At the same time,considering the spread of multiple rumors,a multi-source tracing algorithm based on effective distance is proposed.The detection performance and convergence speed of the algorithm are comprehensively verified through simulation experiments on the estimation of the number of sources,objective function,average error distance,and correct detection probability.3)The issue of traceability of rumors under the anti-rumor mechanism is studied.Introduce anti-rumor mechanism,there are rumor sources and supervision sources in the network.After the supervision source is activated by the rumor,it releases rumor information to achieve the purpose of curbing further spread of the rumor.Under the effect of anti-rumor mechanism,based on SI propagation model,combined with community characteristics and response time of supervision source,a rumor source estimator of anti-rumor mechanism is constructed.Finally,the influence of anti-rumor mechanism on rumor diffusion is analyzed through simulation experiments,and the performance of the estimator is verified from the aspects of correct detection probability and error distance.