Automated LLM-Based Accessibility Remediation: From Conventional Websites to Angular Single-Page Applications

2026-02-19

AI Summary

This work addresses the pronounced challenges of web accessibility in dynamic single-page applications (SPAs), where traditional manual remediation is inefficient and costly. It presents the first approach to leverage large language models (LLMs) for automated accessibility repair in SPAs, introducing a unified, modular workflow that integrates DOM manipulation, source code analysis, and accessibility auditing tools. The proposed framework achieves repair rates of 80% on static websites and 86% on Angular-based SPAs, while also generating context-aware semantic descriptions for images. Designed to preserve visual consistency and system stability, the framework enables end-to-end automated accessibility remediation across both static and dynamic frontends.

Technology Category

• Humans and AI: AI for Accessibility
• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Web accessibility remains an unresolved issue for a large part of the web content. There are many tools to detect errors automatically, but fixing those issues is still mostly a manual, slow, and costly process in which it is easy for developers to overlook specific details. The situation becomes even more complex with modern Single-Page Applications (SPAs), whose dynamic nature makes traditional static analysis approaches inadequate. This work proposes a system that aims to address this challenge by using Large Language Models (LLMs) to automate accessibility fixes. The proposal presents a modular workflow applicable to both static websites and complex Angular projects. The framework actively implements corrections within the DOM of static web pages or the source code of SPAs. The system was tested on 12 static websites and 6 open-source Angular projects, fixing 80% of the accessibility issues on public websites and 86% of the issues on Angular applications. Our proposal also generates meaningful visual descriptions for images while preserving the application's design and stability. This work contributes to ensuring that accessibility stops being a technical debt deferred to the future and becomes a natural part of everyday development workflows.

Problem

• web accessibility
• Single-Page Applications
• accessibility remediation
• dynamic web content
• technical debt

Innovation

• Large Language Models
• Web Accessibility
• Single-Page Applications
• Automated Remediation
• Angular