2025-03-122025-03-122024LINS, Carolina Santos. Classificadores supervisionados para relatos em boletins de ocorrências policiais. Orientador: Filipe Saraiva. 2024. 21 f. Trabalho de Curso (Bacharelado em Ciência da Computação) – Faculdade de Computação, Instituto de Ciências Exatas e Naturais, Universidade Federal do Pará, Belém, 2024. Disponível em: https://bdm.ufpa.br/jspui/handle/prefix/7834. Acesso em:.https://bdm.ufpa.br/jspui/handle/prefix/7834Due to the significant incidence of property crimes consistently re corded in police reports, databases of public security accumulate a substantial number of such cases. One of the responsibilities of the State of Par´a’s Sub Secretariat of Intelligence and Criminal Analysis, an entity subordinate to the Secretary of Public Security and Social Defense of the State of Par´a, is the clas sification of these occurrences, aiming not only for statistical purposes but also to provide support for decision-making related to public security. This work describes the process of developing supervised classifiers, encompassing the methodology employed to handle extremely imbalanced data, a common cha racteristic in datasets related to criminal incidents. Three distinct classifiers were built: one to identify the location, another to classify the victim, and the last one to identify the items and the modus operandi. They achieved an accu racy of 75,86% 90,33% and 80,56% respectively.Acesso AbertoBase de dados de segurança públicaBoletim de ocorrências policiaisEstatísticaTomada de decisõesAnálise automatizada do conteúdo narrativoCNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAOTrabalho de Curso - Graduação - Monografia