De Chamados de Suporte a Conhecimento Operacional: uma Ferramenta de Recomendação Semântica no Governo Digital

• Ana Clara Dias Leal UFSJ
• Lucas Daniel Lana Maciel UFSJ
• Antonio Fiuza de Sousa MGI
• Vanessa Montiel MGI
• Elisa Tuler de Albergaria UFSJ / MGI
• Leonardo Chaves Dutra da Rocha UFSJ

DOI: https://doi.org/10.5753/webmedia_estendido.2026.34280

Resumo

Government support services accumulate large volumes of free-text tickets documenting recurring problems, user questions, procedures, and applied solutions. However, their unstructured nature makes large-scale analysis and reuse difficult. This paper presents an interactive, human-in-the-loop tool to support ticket analysis, triage, and response reuse in government information systems. Built on a real dataset of approximately 40,000 tickets provided by the Brazilian Ministry of Management and Innovation in Public Services, the tool integrates natural language processing, topic organization, semantic recommendation, local language model support, and interactive visualization. It combines two modules: a semantic recommender that retrieves similar historical tickets and previous answers for a new query, and a managerial dashboard that highlights interpretable topics, recurrent bottlenecks, and temporal patterns. A qualitative evaluation with domain specialists indicates that the tool improves access to dispersed historical knowledge, supports response reuse, and provides managers with a consolidated view of operational issues.

Palavras-chave: Digital Government, Support Tickets, Interactive Dashboard

Referências

Daniel Carvalho, Antônio Pereira, Elisa Tuler, Diego Roberto Colombo Dias, Washington Cunha, and Leonardo Rocha. 2025. Uma Ferramenta para Organização Semântica de Informações de Helpdesk por meio de Modelagem de Tópicos. In Anais Estendidos do XXXI WebMedia. DOI: 10.5753/webmedia_estendido.2025.13896

Maarten Grootendorst. 2022. BERTopic: Neural Topic Modeling with a Class-Based TF-IDF Procedure. arXiv preprint arXiv:2203.05794 (2022). DOI: 10.48550/arXiv.2203.05794

Jeff Johnson, Matthijs Douze, and Hervé Jégou. 2017. Billion-scale similarity search with GPUs. arXiv arXiv:1702.08734 (2017).

Gabriel Prenassi, Ana Machado, Davi Freitas, André Lima, Raquel O. Prates, Antonio Landim, Leonardo Rocha, and Elisa Tuler. 2025. Enhanced Analysis of User Perceptions Through Natural Language Processing Approaches. In INTERACT 2025. Vol. 16110. 453–476. DOI: 10.1007/978-3-032-05005-2_23