Multi-Source Detection Based on Overlapping Community Partition and Nonbacktracking Matrix MSI in Social Network
The evolution of online social networks has greatly facilitated the dissemination of information.However,concurrently,it has become a significant channel for the spread of rumors and other false information.The detection of information sources within social networks plays a pivotal role in controlling information propagation.To solve the issue of insufficient accuracy in the existing methods of multi-source detection in social networks,a novel multi-source synchro-nous detection method,named SVT-BiasMSI,has been proposed.This approach transforms the complex task of multi-source detection into several simpler single-source detection tasks.It leverages the Jaccard coefficient of local degree central nodes and their neighboring nodes to select seed nodes.Following this,the Voronoi method is used for preliminary community partitioning within the social network.The tolerance neighborhood method is then implemented to identify nodes in the overlapping sections between partitions,thereby facilitating the division of overlapping communities.In order to balance the consideration of both global and local network information,the SVT-BiasMSI method integrates an improve-ment of the nonbacktracking matrix multi-source detection method based on contagion neighborhood bias,thereby enhancing the precision of single-source detection.Experiments conducted on various social network datasets demon-strate that the proposed method significantly improves the precision of multi-source detection and localization in online social networks.
social networksmulti-source detectionoverlapping community partitionnonbacktracking matrixconta-gion neighborhood bias
万方数据
维普
