Research on Government Web Accessibility Automated Auditing Technology Based on Static Analysis and DOM Rule Engine

Abstract

Aiming at the industry pain‑points of high labour costs for manual compliance auditing of domestic government Web accessibility, dynamic DOM omission detection, incomplete static source‑code parsing, poor adaptability of general‑purpose rules to specific scenarios, and high false‑positive rates of mainstream detection tools, this paper proposes an automated auditing framework for government Web accessibility that integrates static code analysis and a DOM rule engine. This paper constructs a layered rule ontology library compatible with the WCAG 2.1‑AA international standard and domestic government accessibility specifications, designs a DOM rule engine with multi‑node association constraint reasoning capability, and builds a multi‑source static‑analysis pipeline covering HTML, CSS and JavaScript. A defect‑entity alignment algorithm and a weighted confidence fusion model are adopted to eliminate detection conflicts between the two modules, so as to output graded and standardised auditing results. System verification is performed relying on a ground‑truth dataset of 427 manually annotated government webpages (Fischer T, Lundell B & Gamalielsson J., 2025). The experimental results show that compared with mainstream tools such as axe‑core and Lighthouse, the precision, recall and F1‑score of the proposed method reach 87.2%, 84.5% and 85.8% respectively. Its F1‑score is 9.7% higher than that of axe‑core, and its false‑positive rate decreases by 11.3% (Gunderson J & Schepers D., 2021). The proposed method can accurately identify accessibility defects of government‑specific components such as government forms, pop‑up windows and data tables, and provides reliable technical support for compliance construction and inclusive transformation of government‑website accessibility.