| Monitoring anomalies in social networks has been a hot topic in the field of social network analysis.Due to the diversity and complexity of social network data,the use of control charts to monitor and alert social networks can provide more convenience to people’s life.In the past,the monitoring of social networks was usually considered for the whole network.However,in real life,the target team to be monitored is often a part of the participants in the network.If we monitor the social network reasonably and effectively,we can detect the trend of some malicious events in the social network as early as possible and propose targeted solutions to protect people’s life and property.Based on this background,this paper has done the following four aspects:（1）When the target team is known,a Poisson distribution of counted data in a dynamic social network is investigated,and three scenarios are discussed: the target team is a collaborative team,the master leader team,and the entire network,and a monitoring strategy based on the CUSUM control chart is proposed.（2）When the target team is unknown,this paper proposes an efficient search method to estimate the set of candidate teams for collaborative teams and master leader teams,respectively.This reduces the computational complexity when using the exhaustive method to search for target teams.（3）Based on the analysis of the structure of social networks,it is clear that in dynamic heterogeneous networks,the expected communication counts among participants may change over time and with individual pairs,while in homogeneous networks they are constant.Therefore,monitoring strategies for both homogeneous and heterogeneous networks are also considered.（4）The proposed method is simulated and found to be effective in monitoring communication bursts among participants in dynamic social networks.The proposed monitoring method is also used for the detection and alerting of spoofing events in group video sessions. |
PDF Full Text Request