Caraterização das redes de infratores extraídas de ocorrências policiais e identificação de pessoas-chave

Abstract

Police of Minas Gerais. Several empirical measurements were made to characterize the network and extract important information concerning it's operation and how the individuals are related. The first part consisted in the preparation of the base, aiming to extract the individuals stored in that base and point out the replicas in order to identify the individuals uniquely in that base. After carrying out the deduplication of the base, it was possible to extract the criminal networks which was obtained in such a way that two offenders are related if they have appeared in any police report together. Analysis showed that the criminals' networks can be classified as small-world which indicates that the flow of information in these networks works well and that communication is done efficiently. Thus, identifying individuals who play important roles in these networks and isolate them can help to dismantle these criminal organizations and therefore these individuals should be the targets of police actions. In this work, we also compared different algorithms used in the identification of important nodes. It was possible to show that the use of the metric of betweenness is the most appropriate because it leads to an effective result, very close to the optimum, but it is computationally more efficient than the optimal solution. We hope that the results obtained here and the tool for identifying key people may assist in the police's investigative work by helping them solve cases faster and disrupt the organizations more efficiently.