Exploring Determinant Factors in the Archiving of Criminal Investigations in Brazil Through LLMs, RAG, Prompt Engineering, and Data Science

• Grant number: 25/04538-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: May 01, 2025
• End date: April 30, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science
• Principal Investigator: Antonio Castelo Filho
• Grantee: Gustavo Cardozo de Moraes Moreira
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 13/07375-0 - CeMEAI - Center for Mathematical Sciences Applied to Industry, AP.CEPID

Abstract

This project aims to investigate the factors that affect the archiving of criminal investigations in Brazil, focusing on crimes against life. To achieve this, a database from the Public Consultation System of the Public Prosecutor's Office of the State of São Paulo (MPSP) will be used, enabling the extraction and structuring of relevant case information.The planned approach involves applying Natural Language Processing (NLP) techniques, with the development of a pipeline based on Retrieval-Augmented Generation (RAG) to organize and structure the extracted data and validate the models using statistical metrics. With the structured data, we will apply advanced Data Science techniques to identify patterns and factors that impact the archiving of investigations.Therefore, we expect that the findings will contribute to a better understanding of the challenges faced by the criminal justice system, helping to improve the investigation strategies, and enhance the use of NLP and Data Science approaches in the legal context. This approach will also foster the student's development, expanding their academic knowledge beyond the core curriculum.