Law enforcement and intelligence agencies worldwide struggle to find effective ways to fight organized crime and reduce criminality. However, illegal networks operate outside the law and much of the data collected is classified. Therefore, little is known about the structure, topological weaknesses, and control of criminal networks. We fill this gap by presenting a unique criminal intelligence network built directly by the Brazilian Federal Police for intelligence and investigative purposes. We study its structure, its response to different attack strategies, and its structural controllability. Surprisingly, the network composed of individuals involved in multiple crimes of federal jurisdiction in Brazil has a giant component enclosing more than half of all its edges. We focus on the largest connected cluster of this network and show it has many social network features, such as small-worldness and heavy-tail degree distribution. However, it is less dense and less efficient than typical social networks. The giant component also shows a high degree cutoff that is associated with the lack of trust among individuals belonging to clandestine networks. The giant component of the network is also highly modular (Q=0.96) and thence fragile to module-based attacks. The targets in such attacks, i.e. the nodes connecting distinct communities, may be interpreted as individuals with bridging clandestine activities such as accountants, lawyers, or money changers. The network can be disrupted by the removal of approximately 2% of either its nodes or edges, the negligible difference between both approaches being due to low graph density. Finally, we show that 20% of driver nodes can control dynamic variables acting on the whole network, suggesting that non-repressive strategies such as access to basic education or sanitation can be effective in reducing criminality by changing the perception of driver individuals to norm compliance.Electronic supplementary materialThe online version of this article (10.1007/s41109-018-0092-1) contains supplementary material, which is available to authorized users.

The networked nature of criminals using the dark web is poorly understood and infrequently studied, mostly due to a lack of data. Rarer still are studies on the topological effectiveness of police interventions. Between 2014 and 2016, the Brazilian Federal Police raided a child pornography ring acting inside the dark web. With these data, we build a topic-view network and compare network disruption strategies with the real police work. Only 7.4% of the forum users share relevant content, and the topological features of this core differ markedly from other clandestine networks. Approximately 60% of the core users need to be targeted to fully break the network connectivity, while the real effect of the arrests was similar to random failure. Despite this topological robustness, the overall "viewership network" was still well disrupted by the arrests, because only 10 users contributed to almost 1/3 of the total post views and 8 of these were apprehended. Moreover, the users who were arrested provided a total of 60% of the viewed content. These results indicate that for similar online systems, aiming at the users that concentrate the views may lead to more efficient police interventions than focusing on the overall connectivity. In general, internet content is considered as belonging to either the surface web or the deep web depending on the possibility of regular search engines to index it. The set of web pages that for any reason are not accessible by web crawlers is typically called the deep web (e.g., subscription information, medical records, financial records, government resources). Due to this property, it is very hard to estimate the relative size of both layers, however, it is widely believed that the deep web is much larger than the surface web. In an even deeper layer, there is an underlying web, called the dark web, in which there is a significant effort to keep users and their data anonymous, by requiring specific configuration methods and software 1. Precisely because of this feature, the dark web is sometimes used for illegal purposes that range from the black market to fraud services, to child pornography and terrorism. One of the most well-known platforms on the dark web is the Tor network (there are other services such as Freenet and I2P) 2 , an anonymity network originally developed by DARPA in the mid-1990s and lately made available worldwide 3. Due to its high level of anonymity, the Tor network has been used by criminals to run illegal sites and forums, some of which have been seized by law enforcement agencies worldwide 4. From 2014 to 2016, Brazilian Federal Police Agents monitored the activities of individuals forming a pedophile forum on the Tor project during the so-called Operation Darknet 5. The investigation lasted for 2 years and resulted in the identification of 182 targets (out of almost 10,000 users), several prison sentences, search warrants and in the rescue of at least 6 children that were being abused 6,7. After that stage, the monitored message board was deactivated by a court or...

In the multidisciplinary field of Network Science, optimization of procedures for efficiently breaking complex networks is attracting much attention from a practical point of view. In this contribution, we present a module-based method to efficiently fragment complex networks. The procedure firstly identifies topological communities through which the network can be represented using a well established heuristic algorithm of community finding. Then only the nodes that participate of inter-community links are removed in descending order of their betweenness centrality. We illustrate the method by applying it to a variety of examples in the social, infrastructure, and biological fields. It is shown that the module-based approach always outperforms targeted attacks to vertices based on node degree or betweenness centrality rankings, with gains in efficiency strongly related to the modularity of the network. Remarkably, in the US power grid case, by deleting 3% of the nodes, the proposed method breaks the original network in fragments which are twenty times smaller in size than the fragments left by betweenness-based attack.

Corruption crimes demand highly coordinated actions among criminal agents to succeed. But research dedicated to corruption networks is still in its infancy and indeed little is known about the properties of these networks. Here we present a comprehensive investigation of corruption networks related to political scandals in Spain and Brazil over nearly three decades. We show that corruption networks of both countries share universal structural and dynamical properties, including similar degree distributions, clustering and assortativity coefficients, modular structure, and a growth process that is marked by the coalescence of network components due to a few recidivist criminals. We propose a simple model that not only reproduces these empirical properties but reveals also that corruption networks operate near a critical recidivism rate below which the network is entirely fragmented and above which it is overly connected. Our research thus indicates that actions focused on decreasing corruption recidivism may substantially mitigate this type of organized crime.

No abstract

We present a model for network transformation mediated by confinement, as a demonstration of a simple network dynamics that has a direct connection with real world quantities. The model has the capacity of generating complex structures similar to real world networks by the use of two parameters. Starting from an Erdös-Rényi network, nodes are randomly selected to be temporarily confined. Confined nodes form new links among themselves at the same pace they lose connections with the outside nodes. As the network evolves according to the parameters of the model, a series of non trivial network characteristics emerge: the formation of stable heterogeneous degree distributions similar to those of empirical networks, an increasing clustering coefficient, and the emergence of communities outside the confined space. Different from the traditional benchmarks used to create modular networks, there is no arbitrary definition of the number of modules, nor node meta-data defining it as a member of a particular community, nor a tunable parameter directly related with expected modularity. Modules emerge as a result of the dynamics while nodes move among them as connections are rewired. The proposed algorithm has the potential to simulate community dynamics cases in situations where time stamped network data is scarce or absent.

Abordam-se neste artigo a origem, comportamento e características topológicas das facções criminosas brasileiras segundo o prisma da ciência de redes. Mostra-se a estreita relação deste fenômeno com a dinâmica topológica de confinamento, resultando no próprio Estado como catalisador da gênese faccional. Se por um lado a organização da teia criminal brasileira é menos hierarquizada e mais horizontal que outros agrupamentos criminosos típicos como as máfias italianas e as células terroristas, por outro lado, esta ordem organizacional expõe fragilidades que podem ser exploradas pelo sistema de controle criminal. Argumenta-se, pois, que a neutralização seletiva de Alvos Topológicos de Alto Retorno (ATAR) em Regime Disciplinar Diferenciado Pleno (RDD Pleno) tem o condão de fragmentar a rede complexa de facções criminosas nacionais, causando uma forte e consistente redução nos índices de violência. A identificação de ATAR se mostra como tarefa híbrida, pois não prescinde das ferramentas tradicionais de investigação, bem como de métodos matemáticos próprios.

Recent research has shown that criminal networks have complex organizational structures, but whether this can be used to predict static and dynamic properties of criminal networks remains little explored. Here, by combining graph representation learning and machine learning methods, we show that structural properties of political corruption, police intelligence, and money laundering networks can be used to recover missing criminal partnerships, distinguish among different types of criminal and legal associations, as well as predict the total amount of money exchanged among criminal agents, all with outstanding accuracy. We also show that our approach can anticipate future criminal associations during the dynamic growth of corruption networks with significant accuracy. Thus, similar to evidence found at crime scenes, we conclude that structural patterns of criminal networks carry crucial information about illegal activities, which allows machine learning methods to predict missing information and even anticipate future criminal behavior.