Computational intelligence techniques for mining cyber criminal network from online social media

Abstract

Social Network Analysis is a methodology for analysing the social networks to infer useful information for decision making. In recent days, the social network data comes from different domains like crime events, opinions of users, friendship association. These data are available in online social media like newsletters, Facebook, twitters, blogs and others. Criminal Network Analysis is the sub set of Social Network Analysis which focuses on the analysis of criminal networks over the online social media/ collected data. But the crime data are covert and provides a dark network structure which hides information related to the network structure in computer-based analysis system. Hence careful analysis of crime network data has a significant impact on the results of analysis. Community Detection and Prominent Person Identification are the two major tasks considered in this research work for analyzing and extracting knowledge from the bipartite structured crime network data. An important issue in the analysis of crime network data is based on the bipartite structure. The two modes in the network have to be considered for analysing, because both play a vital role in the inferences done. This research work proposes a novel idea for considering the multi-mode structure with single criteria and multiple criteria-based crime network in a single point of time for Community Detection and Prominent Person Identification. The proposed Modified Cluster Walktrap (MCWT) approach for Community Detection in Single Criteria based multi-mode crime network uses the probability-based measure to evaluate group of persons involved in common crime and group of crime done by common people by avoiding the difficulty of adjacency matrix. newline