Identifying criminal organizations from their social network structures

Abstract

Identification of criminal structures within very large social networks is an essential security feat. By identifying such structures, it may be possible to track, neutralize, and terminate the corresponding criminal organizations before they act. We evaluate the effectiveness of three different methods for classifying an unknown network as terrorist, cocaine, or noncriminal. We consider three methods for the identification of network types: evaluating common social network analysis metrics, modeling with a decision tree, and network motif frequency analysis. The empirical resultsshow that these three methods can provide significant improvements in distinguishing all three network types. We show that these methods are viable enough to be used as supporting evidence by security forces in their fight against criminal organizations operating on social networks.

An Empirical Study of the Perception of Criminality through Analysis of Newspapers Online

“…The state of the art indicates the adoption of technologies, the exploration of information, and the impact of these on police practices on the ground, through the organizational and cultural integration of crime analysis, the technological support of analytical practices and the incorporation crime analysis for police practices (Sanders and Condon, 2017), inclusion of geographic features in machine learning algorithms for the prediction of crime based on the network by incorporating a criminal rack (Lin et al, 2018), development of algorithms to find patterns of money laundering criminals (Badal-Valero et al, 2018), development of fuzzy clustering algorithms K-means to obtain criminal critical points, indicating locations with high crime incidence, together with formal concept analysis used to extract visual models that describe patterns that characterize criminal activities (De Farias et al, 2018), analysis of criminal network activities through the application of deep reinforcement learning, applied to the development of a prediction model of hidden links of criminal networks (Lim et al, 2019a(Lim et al, , 2019b) and link prediction model that incorporates a merger of metadata with a criminal data set that evolves over time (Lim et al, 2020), fitting statistical models and machine learning for predicting recidivism (Dressel and Farid, 2018;Van Der Heijden, 2009, 2019), the identification of criminal organizations from social network structures, through the evaluation of common metrics for social network analysis, modeling with decision trees and frequency analysis of network motives (Cesur et al 2017;Çinar et al, 2019), detection and prevention of fraudulent activities related to financial institutions (Makki et al, 2019), visual content generation and natural language processing as a method of teaching skills in police academies (De Sousa Netto et al, 2019), serial crime detection by linking the "modus operandi" (M.O.) and the information of the criminal process, using a natural language processing method to extract the characteristics of the action and object of the criminal process, in addition to an information entropy method to weigh the similarity of the action and the characteristics of the object to obtain the comprehensive similarity of the penal process of criminals (Li and Qi, 2019) and detection of the intention of potential criminal acts through social networks through the generation of ontologies (Saldaña et al, 2019;Yang et al, 2008) (mainly of a specific slang) and machine learning techniques (de Mendonça et al, 2020), among others.…”