Título: Generating Personalized Feedback from Professional Profile Tests Using RAG: A Case Study with MAPER
Autores: Erik Jhones Freitas Do Nascimento, Jessyca A. Bessa, Maryane C. Lima, Igor Rafael Silva Valente, Saulo M. Maia & Maria Lúcia Rodrigues Corrêa
Resumo: Professional profile mapping tests play a crucial role in identifying talents and guiding employee development. However, the growing demand for these evaluations has resulted in increased workload and saturation for the professionals conducting them. Large Language Models (LLMs), when properly integrated, offer promising support for automating report generation and diagnostic feedback. In this study, we explore the use of Retrieval-Augmented Generation (RAG) combined with a structured database of psychological feedback to automatically generate personalized reports for the MAPER test. Our method retrieves example feedback based on specific combinations of competencies and scores, then prompts an LLM to generate fluent and semantically rich output. Experimental results demonstrate that our approach achieves up to 77% semantic similarity compared to expert-generated feedback, indicating strong potential for real-world applications.
Palavras-chave: retrieval augmented generation; large language model; professional profile mapping test.
Páginas: 6
Código DOI: 10.21528/CBIC2025-1132688
Artigo em PDF: CBIC_2025_paper1132688.pdf
Arquivo BibTeX:
CBIC_2025_1132688.bib