ECO: desenvolvimento de um sistema inteligente de gestão de demandas para ouvidorias públicas com foco em prazos legais (Lei nº 13.460/2017), LGPD e análise de sentimento

Abstract

Public ombudsman offices are essential channels for social participation and transparency in public administration, rigorously regulated by federal legislation from 2017. However, managing a high volume of complaints under strict deadlines and ensuring privacy compliance with the General Data Protection Law impose severe operational challenges. This work presents the development and implementation of the ECO web system, an intelligent platform designed for the General Ombudsman Office of the Federal Institute of Piauí. The artifact proposes an automated end-to-end management, incorporating a Kanban board for workflow optimization, automated monitoring of service level agreements, and routines for preserving anonymity. Additionally, natural language processing is employed through the BERTimbau model for sentiment analysis of free-text messages received from citizens, automating preliminary screening and urgency detection. Architecturally, the solution adopts the separation of concerns pattern in a monorepo format orchestrated via Docker Compose, utilizing an application programming interface with Django Rest Framework on the main application server, a reactive Next.js interface on the client, and a PostgreSQL database. Technical validation, conducted through automated testing with Pytest and simulations using dummy data, demonstrates the platform’s viability, security, and efficiency, mitigating neural processing bottlenecks using asynchronous queues and ensuring timeliness in citizen service.