Collective Intelligence with Large Language Models for the Review of Public Service Descriptions on Gov.br

Rafael Marconi Ramos, Rafael Marconi Ramos, Pedro Brom, Pedro Brom, João Gabriel de Moraes Souza, João Gabriel de Moraes Souza, Li Weigang, Vinícius Di Oliveira, Vinícius Di Oliveira, Silvia Reis, Silvia Reis, Jose Salm Junior, Jose Salm Junior, Vérica Freitas, Herbert Kimura, Herbert Kimura, Daniel Cajueiro, Daniel Cajueiro, Gladston Luiz da Silva, Victor Celestino, Victor Celestino

2025

Abstract

This paper presents an intelligent multi-agent system to improve clarity, accessibility, and legal compliance of public service descriptions on the Brazilian Gov.br platform. Leveraging large language models (LLMs) like GPT-4, agents with specialized contextual profiles simulate collective deliberation to evaluate, rewrite, and select optimal service texts based on ten linguistic and seven legal criteria. An interactive voting protocol enables consensus-based editorial refinement. Experimental results show the system produces high-quality texts that balance technical accuracy with linguistic simplicity. Implemented as a Mixture of Experts (MoE) architecture through prompt-conditioning and rhetorical configurations within a shared LLM, the approach ensures scalable legal and linguistic compliance. This is among the first MoE applications for institutional text standardization on Gov.br, establishing a state-of-the-art precedent for AI-driven public sector communication.

Paper Citation

in Harvard Style

Ramos R., Brom P., Souza J., Weigang L., Di Oliveira V., Reis S., Salm Junior J., Freitas V., Kimura H., Cajueiro D., Luiz da Silva G. and Celestino V. (2025). Collective Intelligence with Large Language Models for the Review of Public Service Descriptions on Gov.br. In Proceedings of the 21st International Conference on Web Information Systems and Technologies - Volume 1: WEBIST; ISBN 978-989-758-772-6, SciTePress, pages 301-312. DOI: 10.5220/0013831100003985

in Bibtex Style

@conference{webist25,
author={Rafael Ramos and Pedro Brom and João Souza and Li Weigang and Vinícius Di Oliveira and Silvia Reis and Jose Salm Junior and Vérica Freitas and Herbert Kimura and Daniel Cajueiro and Gladston Luiz da Silva and Victor Celestino},
title={Collective Intelligence with Large Language Models for the Review of Public Service Descriptions on Gov.br},
booktitle={Proceedings of the 21st International Conference on Web Information Systems and Technologies - Volume 1: WEBIST},
year={2025},
pages={301-312},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013831100003985},
isbn={978-989-758-772-6},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 21st International Conference on Web Information Systems and Technologies - Volume 1: WEBIST
TI - Collective Intelligence with Large Language Models for the Review of Public Service Descriptions on Gov.br
SN - 978-989-758-772-6
AU - Ramos R.
AU - Brom P.
AU - Souza J.
AU - Weigang L.
AU - Di Oliveira V.
AU - Reis S.
AU - Salm Junior J.
AU - Freitas V.
AU - Kimura H.
AU - Cajueiro D.
AU - Luiz da Silva G.
AU - Celestino V.
PY - 2025
SP - 301
EP - 312
DO - 10.5220/0013831100003985
PB - SciTePress