Criminal analysis approach for the improvement of situational awareness using data and information fusion

Abstract

In the field of risk management based on criminal data, SAW is a crucial factor in revealing trends, incidence of threats and the increase or decrease of imminent risks. A limited SAW can compromise analysts' understanding of what has actually happened and is happening, leading to poor decision making, which can result in disastrous consequences for people, assets or the environment [2]. The process of acquiring SAW is even more challenging since most of the data are provided by human intelligence (HUMINT), as is the case with reports of crimes reported to security service centers (e.g., 190 response service of Military Police of São Paulo), which record the reports in the Occurrence Records. Typically, HUMINT data are incomplete, outdated, inconsistent and sometimes even influenced by cultural factors, influencing the computational processes that process the data and stimulate the analyst's SAW. In this way, Data and Information Fusion techniques were developed to identify a synergy between heterogeneous data and information and to integrate it, providing smaller but more significant information to reduce uncertainties [1, 3, 4, 5, 6, 7,8]. However, particularities of the cited challenges require more robust processes of preprocessing, search for relevant elements and synergetic association. In the context of criminal analysis, it is worth mentioning that the country's criminal databases have been formed recently and have few consistent historical series. In addition, a second problem is the poor quality of the data collected, mainly regarding the reliability of registered addresses as places of crime, due to the inaccuracy of the victim in the place where he was victimized, and the non-prioritization of this data by Part of the collectors (usually civilian police), which focuses more on the description of the event, for legal purposes, than on location data. In addition, most electronic systems for recording occurrences, allows the completion of the record even without the address of the fact or only with a reference. This aspect is particularly important for the decision-making process, since the (appropriate) practice of criminal mapping has become popular in the country, as a resource for data analysis for decision-making. However, by ignoring the poor quality of location information and the absence of data processing routines, maps today do not reflect the actual incidence of crime and crime problems, unless they undergo individual records review processes. The third and final problem is even more serious. Databases do not have structured data for criminal analysis that will inform public policy decisions or private or community crime prevention and control actions. The criminal data are cataloged according to a legal classification, the penal code and the special penal laws, focused on the typification of the aggressor's action and not on the definition of the criminal phenomenon. The criminal problem is always related to a specific environment that "encourages", for the low risk to the criminal, several crimes in the same environment (for example, bag theft, cell phone theft, lightning kidnapping, occurred in the same environment). For the identification of these "criminal problems", the techniques of data fusion, organized from the assumptions of situational criminal analysis and economic theory of crime, will be decisive in this proposal. In order to overcome these challenges, the objective of this work is the development of a fusion process composed of specification, implementation, application of hierarchical and multicriteria fusion algorithms, and the development of a process and system of criminal data analysis as a case study. (AU)