10.29327/1884517.29-29
Accessibility is a key quality requirement for digital systems, especially in the public sector, where inclusive access is essential. However, specifying accessibility requirements remains challenging in Requirements Engineering, as it requires specialized knowledge and careful interpretation of standards such as WCAG. Although Large Language Models (LLMs) have been explored for requirements-related tasks, there is limited empirical evidence of their use for accessibility requirements in real-world projects. This paper reports an industry experience using ChatGPT to support the specification of accessibility requirements for the Plataforma Antonieta de Barros, a public-sector digital platform designed to support data-driven decision-making in education. The goal is to investigate how an LLM can assist in generating requirements aligned with WCAG 2.2 and to evaluate their quality from practitioners’ perspectives. The findings suggest that ChatGPT can support the initial specification of clear, relevant accessibility requirements, but human validation and iterative refinement are essential to ensure quality and applicability. The experience also highlights the importance of structured prompts and multidisciplinary evaluation to transform AI-generated outputs into actionable requirements for real-world software projects.
Keywords: Accessibility Requirements; Large Language Models; Requirements Engineering
@inproceedings{wer202628,
author = {Marques, A. B. and Oliveira, R. and Luna, R. and Sousa, A. and Macêdo, J. A. and Andrade, R.},
title = {Using ChatGPT to Support Accessibility Requirements Specification: Experience from a Public Sector Platform},
booktitle = {Anais do Workshop em Engenharia de Requisitos - Proceedings of the 29th Workshop on Requirements Engineering (WER2026)},
year = {2026},
issn = {2675-0066},
isbn = {978-65-02-19591-8},
doi = {10.29327/1884517.29-29}
}