Um pipeline reprodutível e um dataset aberto para auditoria de conformidade WCAG em design systems públicos
Resumo
This paper presents an open audit workflow and dataset for accessibility analysis of public design system documentation. The instrument combines archived axe-core results, rule-and-selector deduplication, explicit scope reconciliation, component-level probes and an eleven-item manual checklist. Its application to Bold and DSGov illustrates traceability from findings to interface elements. Reprocessing the archived exports yields 119 and 278 unique records; removing enhanced-contrast AAA records yields the historical totals of 75 and 202. These totals describe the study’s operational scope, which also contains best-practice and WCAG 2.2 checks, rather than a WCAG 2.1 AA conformance certificate. We distinguish reproducible processing of archived data from live recollection and human judgment, document differences between tool versions, and disclose remaining inconsistencies in component analysis. Data are licensed under CC BY 4.0 and scripts under MIT. The contribution is the reusable instrument; the underlying empirical study is identified separately.
Referências
Thomas Fischer, Björn Lundell, and Jonas Gamalielsson. 2025. Coverage of web accessibility guidelines provided by automated checking tools. Universal Access in the Information Society 24, 4 (2025), 3615–3637. DOI: 10.1007/s10209-025-01263-x
W3C. 2018. Web Content Accessibility Guidelines (WCAG) 2.1. W3C Recommendation. [link]
W3C. 2023. Web Content Accessibility Guidelines (WCAG) 2.2. [link]
Thiago Kenji Corrêa Xikota. 2025. Acessibilidade digital em design systems públicos: uma auditoria de conformidade do Bold e DSGov. [link] Trabalho de Conclusão de Curso (Bacharelado em Ciências da Computação), INE/CTC, Universidade Federal de Santa Catarina. Repositório institucional UFSC, handle 123456789/270866.
Thiago Kenji Corrêa Xikota. 2026. Dataset: auditoria de acessibilidade Bold vs DSGov. [link] Repositório consultado em 28/09/2026, revisão b94fbdd.
