MutatedSocioAgentSim (MSAS): semisupervised modelling of multiagent simulation to predict and detect the mutation in a camouflaged social network

Authors: KARTHIKA SUBBARAJ, BOSE SUNDAN

Abstract: A social network is a networked structure formed by a set of agents/actors. It describes their interrelationships that facilitate the exchange and flow of resources and information. A camouflaged social network is one such community that influences the underlying structure and the profile of the agents, to cause mutation. The proposed MSAM is a novel system that simulates a multiagent network whose community structure is analyzed to identify the critical agents by studying the mutations caused due to attachment and detachment of agents. The isolation of the tagged agents will demonstrate disruption of information flow, which leads to the dismantling of the camouflaged community and giving scope for a predictive study about near future reconciliation. The proposed system simulates the 9/11 covert network based on the belief matrix and uses the novel density-based link prediction and suite of fragmentation algorithms for predictive community analysis. MSAM is claimed to be an intelligent system as agents perceive the knowledge from the dynamic environment through the belief matrix and further co-evolve as a community upon which semisupervised methodologies are used to predict the critical agents causing serious mutation.

Keywords: Multiagent simulation, camouflaged social network, semisupervised learning, community structure, change detection, fragmentation, 9/11 covert network

Full Text: PDF