Banca de DEFESA: ALMEIDA SIMÕES DOS SANTOS

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : ALMEIDA SIMÕES DOS SANTOS
DATE: 02/03/2026
TIME: 08:30
LOCAL: Link da videochamada: https://meet.google.com/gcy-qurq-nwb
TITLE:

Applying LLMs for Software Requirements Specification Support

KEY WORDS:

Requirements Engineering; ISO 29148; LLM; RAG; Prompt Engineering; Design Science Research; Textual Quality of Requirements.

PAGES: 95
BIG AREA: Ciências Exatas e da Terra
AREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
SPECIALTY: Engenharia de Software
SUMMARY:

Requirements Specification often suffers from ambiguity, inconsistency, and incompleteness, which compromises software verifiability and increases development costs. This study investigates how Large Language Models (LLMs), guided by Prompt Engineering, can systematize the generation of natural language specifications in compliance with the ISO 29148:2018 standard. The research proposes a Specification Assistant structured into four main pillars: (i) assisted requirement generation based on elicitation results; (ii) the application of few-shot prompting incorporating syntactic patterns and quality attributes defined by the standard; (iii) the integration of Retrieval-Augmented Generation (RAG) for contextualized consultation of best practice guides and standard excerpts; and (iv) an iterative refinement process mediated by human supervision to mitigate subjectivity and scope gaps. Following the Design Science Research (DSR) methodology, the artifact was evaluated through comparative experiments involving three groups of undergraduate students, contrasting the Assistant’s performance against a baseline LLM (ChatGPT). The evaluation criteria focused on clarity, consistency, completeness, and the required writing effort. The contributions of this work include: (1) the operationalization of ISO 29148 recommendations into automated checks; (2) a replicable protocol for the use of LLMs in requirements engineering; and (3) empirical evidence regarding the benefits and limitations of automated support in software quality assurance.

COMMITTEE MEMBERS:
Presidente - 2195240 - MARCIA JACYNTHA NUNES RODRIGUES LUCENA
Interno - 1671962 - EDUARDO HENRIQUE DA SILVA ARANHA
Interna - 1754430 - LYRENE FERNANDES DA SILVA
Externo à Instituição - ENYO JOSÉ TAVARES GONÇALVES - UFC