Augment browsing and standard profiling for enhancing web accessibility
Silvia Mirri, Paola Salomoni, Catia Prandi · 2011 · Proceedings of the International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/1969289.1969297
Summary
This paper presents GAPforAPE (GreaseMonkey And Profiling for Accessible Pages Enhancement), a client-side system that automatically transcodes web content to meet individual users' accessibility needs and preferences. The system is built as a Firefox browser extension using the GreaseMonkey scripting framework and is guided by the principle that "one Web content for everyone" is less effective than "the best Web content for each one." GAPforAPE combines two components: a user profiling system based on the IMS ACCLIP (Accessibility for Learner Information Package) standard, and a content transcoding engine that uses JavaScript to modify the HTML DOM, CSS rules, and page scripts at the client side. The profiling system organizes user preferences into five categories — Text, Color, Audio, Visual, and General — allowing users to specify detailed preferences such as text size, color contrast, image handling (replacement with text alternatives, audio descriptions, or subtitles), and link styling. The system operates on a two-layer model inspired by screen readers like JAWS: site-specific scripts are applied when available for recognized web applications, and a default set of generic scripts handles all other pages. This approach addresses limitations of existing content adaptation methods including server-side approaches (requiring host infrastructure), proxy-based solutions (bandwidth intensive), and service-oriented architectures (complex to maintain).
Key findings
The system was evaluated through a case study focused on improving Facebook accessibility for users with disabilities. The researchers involved 16 blind users who described how they navigate Facebook using screen readers, revealing numerous barriers: inaccessible chat, poorly organized heading hierarchies, cyclic navigation links, ambiguous link text, redundant information, unlabeled form elements, and automatic page refreshes that disrupted screen reader navigation. GAPforAPE scripts addressed these issues by labeling text links correctly, removing redundant content, labeling form elements, removing decorative images, reorganizing heading hierarchies, blocking automatic updates (letting users choose when to refresh), and adding WAI-ARIA live regions for accessible chat functionality. The system demonstrated adaptations for different user profiles — blind users received reorganized DOM with proper semantics, while low-vision users got high-contrast color schemes (yellow text on black background) or enlarged text with simplified layouts. The client-side approach proved advantageous because it required no server infrastructure, worked across any website, and allowed users to set preferences once and have them applied automatically across all browsing sessions.
Relevance
This research explores client-side content adaptation as a pragmatic approach to accessibility — particularly relevant when web developers fail to build accessible sites. The philosophy of personalizing content to individual needs rather than offering a single "accessible version" aligns with modern approaches to inclusive design and user preferences. While the specific technology (GreaseMonkey/Firefox extensions) has evolved, the underlying concept of browser-based content transcoding remains relevant through modern browser extensions and user stylesheets. The detailed analysis of Facebook's accessibility barriers for blind users provides a valuable catalogue of common social media accessibility failures that persist across platforms today: poor heading structure, unlabeled interactive elements, automatic content updates disrupting assistive technology, and redundant navigation. The use of the IMS ACCLIP standard for user profiling demonstrates the value of standardized accessibility preference descriptions — a concept that has gained renewed attention with efforts like the W3C's personalization specifications. For practitioners, the work highlights that accessibility is not one-size-fits-all; different users with the same disability type may have very different preferences for how content should be adapted.
Tags: content transcoding · user profiling · browser extensions · social media accessibility · personalization · client-side adaptation · screen readers · low vision
Standards referenced: WAI-ARIA · IMS ACCLIP · CC/PP · UAProf