Published June 3, 2026 | Version v1.0.0

Sonar Municipal: semantic search pipeline for Brazilian municipal bills

Authors/Creators

  • 1. Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo
  • 1. Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo

Description

Source code accompanying the ICMC-USP undergraduate thesis 'Mineração de Dados e Busca Semântica Aplicadas à Análise de Projetos de Lei Municipais' (Sonar Municipal project). Implements the SAPL data collection pipeline, dense-retrieval indexing with E5 and BM25 baselines, and the evaluation tooling for the companion TREC-style benchmark. Supported by the AI4PEP (Artificial Intelligence for Public Health Emergency Preparedness) network. Companion artifacts on the Hugging Face Hub: model thiagoambiel/sonar_municipal_ptt5_ementa2action, dataset thiagoambiel/sonar-municipal-pl-actions, demo thiagoambiel/sonar-municipal-demo. Companion Zenodo records (dataset, model) will be cross-linked once their DOIs are minted; the thesis BDTD-USP handle will be linked as a related identifier once available.

Notes

Definitive release (v1.0). Companion dataset and model DOIs and the BDTD-USP thesis handle will be added as related identifiers via the Zenodo UI once available, without re-minting (Zenodo allows metadata edits).

Files

thiagoambiel/SonarMunicipal-v1.0.0.zip

Files (6.5 MB)

Name Size Download all
md5:770f90e797402d04365e7545b43e803f
6.5 MB Preview Download

Additional details

Related works

Is supplement to
Software: https://github.com/thiagoambiel/SonarMunicipal/tree/v1.0.0 (URL)
Is supplemented by
Dataset: 10.5281/zenodo.20564283 (DOI)
Software: 10.5281/zenodo.20564326 (DOI)