Research On The Method Of Information Traceability In Social Network

Information traceability is a research hotspot in the field of social network analysis.According to the number of source nodes,information traceability can be divided into single source traceability and multi-source traceability.In the aspect of single source traceability,the existing algorithms focus on posterior estimation and ignore the role and value of prior estimation,which leads to low accuracy,large error distance and low accuracy of normalized ranking.In the aspect of multi-source traceability,network partition is a common and effective method,but this kind of algorithm ignores the overlapping problem between different infection partitions,which leads to the low min partition coverage in the divided area,and poor single source tracing performance in the poor divided area.In view of the defects and shortcomings of the above two kinds of traceability algorithms,this thesis mainly completes the following work:This paper proposes a propagation source tracing algorithm based on priori Estimation.Considering that the real source node participates in the infection process for the longest time and tries to infect the neighbor node for the most times,the algorithm thinks that the infection proportion of the neighbor node should be the largest.Therefor the algorithm calculates the prior estimation value of the source node according to the number relationship between the infected node and the uninfected node in the neighbor node,which effectively makes up for the deficiency of the existing traceability algorithm in prior estimation.Experiments on artificial networks and real networks show that the proposed algorithm has higher detection rate,smaller error distance and higher normalized ranking of real source nodes.This thesis proposes a multi-source traceability algorithm based on overlapping community partition.The algorithm solves the problem of overlapping infection partitions by dividing the infected network into overlapping communities,and ascertains the number of real source nodes by the number of detected communities.In the divided community,the propagation source tracing algorithm based on priori estimation is used to obtain multi source tracing results.Experiments on artificial networks and real networks show that the proposed multi source traceability algorithm based on overlapping community detection has the advantages of high detection rate of the number of source nodes,high min partition coverage,short error distance,and so on;moreover,with the change of network,the algorithm shows good stability.In this thesis,we propose two algorithms to solve the traceability problem,which improve some shortcomings of the existing algorithms.The propagation source traceability algorithm based on prior estimation proposed in this thesis solves the problem that the existing algorithms ignore the role and value of prior estimation for traceability;This thesis proposes a multi-source tracing algorithm based on overlapping community partition to solve the problem of overlapping infection areas.Compared with the existing algorithms,the two algorithms perform well.In this thesis,there are 17 figures,8 tables and 78 references.