ANPC Panel: A Hybrid Framework for Legal Information Extraction, Transparency, and Human-on-the-Loop Validation
Resumo
Legal document analysis plays an important role in supporting institutional oversight and public governance; however, manual extraction of information from large document collections is time-consuming and difficult to scale. This paper presents the ANPC Panel, a hybrid framework that combines Large Language Models (LLMs), retrieval-based extraction, and complementary rule-based strategies to transform Civil Non-Prosecution Agreements (ANPCs) into structured representations for institutional monitoring and large-scale analysis. Beyond automated extraction, the system incorporates algorithmic transparency and Human-on-the-Loop validation mechanisms, allowing users to inspect, contextualize, and audit AI-generated outputs. By combining automated information extraction, transparent analytics, and human supervision, the proposed approach contributes to strengthening efficiency, accountability, and evidence-based decision-making in public-sector environments.
Referências
Fernandes, W. P. D., Frajhof, I. Z., da Franca Couto Fernandes de Almeida, G., Rodrigues, A. M. B., Barbosa, S. D. J., Konder, C. N., Nasser, R. B., de Carvalho, G. R., and Lopes, H. C. V. (2022). Extracting value from brazilian court decisions. Information Systems, 106:101965.
Fernandes, W. P. D., Silva, L. J. S., Frajhof, I. Z., da Franca Couto Fernandes de Almeida, G., Konder, C. N., Nasser, R. B., de Carvalho, G. R., Barbosa, S. D. J., and Lopes, H. C. V. (2020). Appellate court modifications extraction for portuguese. Artificial Intelligence and Law, 28(3):327–360.
Gebauer, M. et al. (2023). A human-in-the-loop approach for information extraction from privacy policies under data scarcity.
Gonçalves, J. P. (2021). Desenvolvimento de painéis bi/eproc de apoio à gestão no tribunal de justiça do tocantins. Technical report, Universidade Federal do Tocantins.
Grattafiori, A., Dubey, A., Jauhri, A., and et al. (2024). The llama 3 herd of models.
Justino, A. F. et al. (2025). A business intelligence platform to support legal data management at the tjma.
Júnior, J. B. T. (2021). Considerações sobre o acordo de não persecução cível – anpc. Revista Procuradoria Geral do Estado do Mato Grosso - PGE/MS, 1(17):24.
Li, Z. et al. (2024). Dmqr-rag: Diverse multi-query rewriting for retrieval-augmented generation. arXiv preprint arXiv:2411.13154.
Neto, H. L. (2019). O mapeamento das ações de saúde pelo tribunal de justiça do estado do pará. Revista de Política Judiciária, Gestão e Administração da Justiça, 5(2):1–17.
Pereira, J., Assumpcao, A., Trecenti, J., Airosa, L., Lente, C., Cléto, J., Dobins, G., Nogueira, R., Mitchell, L., and Lotufo, R. (2024). Inacia: Integrating large language models in brazilian audit courts: Opportunities and challenges.
Rea, L. and Parker, R. (2012). Designing and Conducting Survey Research: A Comprehensive Guide. Wiley.
Ruschel, A. J. (2015). Governo eletrônico: Business intelligence para a modernização do judiciário.
Schroeder, N., Jaldi, C., and Zhang, S. (2025). Large language models with human-in-the-loop validation for data extraction.
Team, G., Kamath, A., Ferret, J., Pathak, S., and et al. (2025). Gemma 3 technical report.
Zambrano, G. (2024). Case law as data : Prompt engineering strategies for case outcome extraction with large language models in a zero-shot setting. Law, Technology and Humans, 6(3):80–101.
