Towards Mapping Brazilian Public Comments on Legislative Text

• Ano de defesa: 2024
• Autor(a) principal: Siqueira, Felipe Alves
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
• Palavras-chave em Português: Domínio legislativo; Legislative domain; Mapeamento em textos; Mapeamento semântico; Modelos de sentenças; Segmentação de texto; Semantic mapping; Sentence models; Text mapping; Text segmentation
• Link de acesso: https://www.teses.usp.br/teses/disponiveis/55/55134/tde-06012025-114009/

This work presents key resources for managing large-scale data generated by digital public participation in the Brazilian legislative process. These resources were developed with the guiding task of mapping public comments to semantically related regions of legislative bills. The resources include machine learning models for legislative text segmentation, a large-scale corpus for the Brazilian legal and governmental domain, in-domain semantic sentence models, and a benchmark for assessing Brazilian legal sentence models. Although these resources were primarily developed for the semantic mapping of public comments to legislative bill regions, they are publicly available and can be used independently in future research.