Defesa – Tese de Guilherme Rego Rockembach
Aluno(a): Guilherme Rego Rockembach
Orientador(a): Lucinéia Heloisa Thom
Título: Proficiency Modeling and Skill Extraction in BPM Education: A Generative AI-Supported Framework
Linha de Pesquisa: Engenharia de Software
Data: 19/06/2026
Hora: 09:00
Local: Esta banca ocorrerá de forma remota. Acesso público disponibilizado pelo link https://mconf.ufrgs.br/webconf/00111501.
Banca Examinadora:
-Evellin Cristine Souza Cardoso (UFG)
-Cristiano Galafassi (Unipampa)
-Dennis Giovani Balreira (UFRGS)
Presidente da Banca: Lucinéia Heloisa Thom
Resumo: Business Process Management (BPM) is a strategic discipline. However, the training of qualified professionals in BPM faces challenges due to the complexity of the subject and the lack of consensus on teaching and assessment methodologies, creating a gap between theory and practice. The literature identifies two central gaps: (i) the lack of a structured definition of the competencies and skills that should be developed in BPM education; and (ii) the absence of a proficiency architecture that describes the progressive development of these competencies, based on cognitive taxonomies and learning models, capable of guiding consistent assessment practices. To directly address these two fundamental gaps, this thesis develops a comprehensive framework supported by Generative Artificial Intelligence for BPM education in higher education. Filling the first gap, the model extracts and structures a clear set of competencies and skills, derived from structured domain-specific content. To resolve the second, it establishes a mathematically grounded proficiency architecture for each stage of learning, providing formal support for diagnostic, formative, and summative assessments powered by Intelligent Tutoring Systems. Conducted under the Design Science Research paradigm, the research developed a computational artifact that utilizes a pedagogical heuristic and Large Language Models to derive and structure educational skills. The generated set was empirically validated by an expert panel with high agreement rates, and its cross-domain generalization capability was proven. The proficiency architecture was validated through a retrospective study of logs in a Virtual Learning Environment. As a contribution, the thesis delivers a reproducible methodological framework that guides pedagogical planning and assessment practices in BPM. Furthermore, the approach is proven to be generalizable, serving as a rigorous methodological template for other disciplines and interdisciplinary areas that lack consolidated benchmarks for adopting AI in education.
Palavras-Chave: Business Process Management. Learning Assessment. Proficiency Framework. Competency-Based Education. Generative Artificial Intelligence.