Using Ontologies as a Foundation for Web Accessibility Tools
Jens Pelzetter · 2018 · Proceedings of the 15th International Web for All Conference (W4A 2018) · doi:10.1145/3192714.3196316
Summary
This extended abstract proposes using formal ontologies — machine-readable knowledge representations using languages like OWL (Web Ontology Language) — as the foundation for web accessibility tools, addressing two major problems: supporting developers in creating accessible websites and remediating the large number of existing inaccessible sites. The motivation is that current accessibility tools are limited: automated testing can only check a subset of WCAG requirements, manual testing reliability depends heavily on tester experience (inexperienced testers find either too few or too many problems), and the rules for accessible web pages are not available in a machine-readable form that tools can reason about programmatically. The paper proposes creating a machine-readable version of the WCAG 2.0 standard and its companion documents using ontologies, enabling tools to not only check compliance but also explain why a rule exists and how to fix violations. Prior work has used ontologies in accessibility contexts: the ABBA project created ontological web page models for a specialised browser providing navigation assistance to visually impaired users; an accessibility assessment environment using ontologies was described in related work; and a combined ontology of WCAG and disabilities was created to help developers understand not just the guidelines but why they matter. In other disability assistance domains, ontologies have been successfully applied to mobility assistance. The approach envisions two types of tools built on the ontology: developer-facing tools integrated into CMS and authoring environments that provide real-time guidance during content creation, and remediation tools that apply client-side refactorings (small JavaScript modifications) to existing inaccessible pages.
Key findings
The proposed ontology would encode three layers of knowledge: (1) the formal WCAG rules and success criteria in a machine-processable format, enabling automated reasoning about which rules apply to a given HTML element or page structure; (2) the rationale behind each rule — which disabilities it addresses and why the requirement exists, connecting technical guidelines to human impact; and (3) remediation techniques, linking each violation to specific fixes that can be applied automatically or semi-automatically. For developer tools, the ontology would enable CMS systems to provide contextual accessibility guidance as content is being created — for example, prompting for alt text when an image is uploaded, or warning about colour contrast issues during theme customisation. For client-side refactoring tools, the ontology would help determine which JavaScript modifications should be applied to a given page to improve its accessibility without requiring changes from the site owner. The author draws a parallel to the ARIA specification, which annotates HTML with additional semantic information for assistive technologies, arguing that ontologies extend this concept to the entire accessibility knowledge domain. The paper acknowledges that creating a comprehensive ontology of WCAG is a substantial undertaking and positions this as doctoral research in progress.
Relevance
This paper addresses a persistent challenge in web accessibility tooling: the gap between the human-readable WCAG specification and what automated tools can reason about. Current automated tools encode accessibility rules as hard-coded checks, making them difficult to update, extend, or explain. An ontological approach would enable tools that not only identify violations but explain their impact on specific user groups and suggest contextual fixes — moving from "this image lacks alt text" to "this image lacks alt text, which means a blind screen reader user cannot understand this product listing, and here is how to write effective alt text for product images." For accessibility practitioners, this vision connects to the broader challenge of making accessibility knowledge actionable for developers who are not accessibility specialists. The German regulatory context (BITV, mandating accessible websites for all federal institutions) provides practical motivation, as government organisations need scalable tools to evaluate and remediate large numbers of pages. As a 2-page extended abstract describing planned doctoral research, this is a conceptual proposal without implementation or evaluation, but the approach of formalising accessibility knowledge as linked, machine-readable data remains a valuable direction for the field.
Tags: web accessibility · ontologies · WCAG · knowledge representation · automated testing · semantic web · OWL · accessibility evaluation · developer tools
Standards referenced: WCAG 2.0 · WAI-ARIA · BITV