Detecting covert members : Quadrant approach for classification and identification of smart criminals

Abstract

Telecommunication metadata is a resourceful tool that can be employed in fighting incessant crimes. One peculiar challenge in this resourceful material is the inability to access the personal status of criminal syndicate. It weakens evidence for identifying conspirators in crimes. Recently, quadrant approach has proven to be a better approach for analyzing metadata to unveil the level of involvement of the direct perpetrators. However, it was found that combination of only degree and betweenness centrality in this approach limits its ability to uncover the key-players, especially distant influencers. In this work, we have proposed an enhanced quadrant approach as a robust method for classifying criminal activities based on their relationships. The method reserves a portion of quadrants for identifying smart(hidden) criminals or fugitives. In order to ensure that all vital covert members are adequately uncovered, this study incorporates more social network analysis (SNA) metrics into the quadrant approach. It was found that using closeness centrality as a feature for strategic positioning make more conspirators become prominent than using betweenness centrality. Thus, the number of detected covert members was increased by detection of smart criminals in the quadrant. Out of four cases that were studied, the case where closeness centrality was combined with degree centrality detected 19.05% of conspirators among the criminal group as against the 2.38% of conspirators that were identified when betweenness and degree centrality were combined. The use of closeness centrality also reduced the inconspicuous conspirators to 59.52%.