Hybrid methodology for analysis of structured and unstructured data to support decision-making in public security

This work proposes a hybrid methodology that enables the integration of structured and unstructured data to support the decision-making process in public security contexts. The proposed methodology facilitates classification and prediction of crime in a given region, making it possible to identify actions to improve public security based on the results. The integration of the data takes place in two main steps: (1) loading and analyzing structured data made available by government agencies; and (2) absorbing, classifying, and analyzing unstructured data from digital platforms such as Twitter, Where I Was Robbed, and CityCop. In this way, it becomes possible to transform these unstructured data into structured data to be incorporated into a historical database on which algorithms can act to classify, measure, and predict crime. To illustrate the applicability of this methodology, we conducted a study in the city of Recife, Brazil. Structured and unstructured data were gathered in order to conduct a neighborhood classification analysis of crime hot spots. Based on that analysis, we conducted a series of actions intended to bring improvements to the region by the local police. We obtained an increase in the algorithms' accuracy rate of 80%, indicating that public security organizations can base their actions on the results of the proposed methodology.