Developing a Semantic User and Device Modeling Framework That Supports UI Adaptability of Web 2.0 Applications for People with Special Needs
Philip Ackermann, Carlos A. Velasco, Christopher Power · 2012 · Proceedings of the International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/2207016.2207018
Summary
This paper presents a semantic modeling framework for dynamically adapting web 2.0 application interfaces based on user preferences and device capabilities, developed within the EU-funded I2Web (Inclusive Future-Internet Web Services) project. The framework addresses a growing complexity problem: as the web evolved from static pages to rich interactive applications with mashups and user-generated content, traditional accessibility approaches (treating the web developer as the sole responsible party) became insufficient. The architecture covers the full lifecycle from authoring to delivery, with four interconnected models. The Web Technology model provides a decoupled vocabulary for describing device capabilities at different granularities — from HTML elements and attributes to CSS properties and images. The Device model describes hardware (display, keyboards, sound cards, assistive technology hardware like Braille lines), software platform (operating system, installed software), and interaction software (screen readers, on-screen keyboards). The User model maps between device capabilities and user preferences through four transformation types: suppression (removing irrelevant UI components), insertion (adding new components like subtitles for DHH users or help cues for older adults), substitution (replacing components, e.g., alt text for images for blind users), and reorganization (simplifying spatial or temporal layouts to reduce cognitive load). All models are built on the W3C Composite Capability/Preference Profiles (CC/PP) 2.0 RDF framework, enabling semantic matching without heavyweight OWL ontology processing.
Key findings
The framework was implemented and tested in three real-world application contexts: a multiplatform e-banking framework, a multiplatform e-government application, and a multimedia content management system. The four-transformation decomposition (suppression, insertion, substitution, reorganization) provides a practical vocabulary for describing any UI adaptation for accessibility. The system supports both single-step transformations (applying all three models at once) and multi-step pipeline transformations (sequential adaptations, similar to Unix pipes and filters). A key architectural decision was using CC/PP rather than full OWL ontologies — previous projects like ACCESSIBLE and AEGIS had built complex ontological user models that became too demanding for management and runtime processing. The CC/PP approach allows two-level hierarchical profiles with components and attributes that can be included directly or referenced, enabling efficient processing. The framework also integrates with remote accessibility evaluation web services using the EARL (Evaluation and Report Language) format, allowing accessibility experts to evaluate applications and feed results into the adaptation process. Initial user testing showed positive results, though the paper notes that further evaluation and generalization of adaptation rules is needed.
Relevance
This work represents an important approach to accessibility that goes beyond static WCAG compliance toward dynamic, personalized interfaces that adapt to individual users' capabilities and devices. The four-transformation taxonomy (suppress, insert, substitute, reorganize) provides a useful mental model for thinking about any accessibility adaptation — from simple alt text (substitution) to simplified layouts for cognitive accessibility (reorganization) to adding captions for video content (insertion). While the specific CC/PP and RDF technologies did not achieve widespread adoption, the underlying concept of model-driven UI adaptation is increasingly relevant as the web moves toward diverse delivery contexts (mobile, wearables, voice, AR/VR) where a single static interface cannot serve all users. Modern approaches like WAI-Adapt (formerly Personalization Semantics) and GPII (Global Public Inclusive Infrastructure) continue this line of research. For practitioners, the paper reinforces that true accessibility in complex web applications requires thinking beyond markup compliance to how interfaces can dynamically reshape themselves based on user needs.
Tags: adaptive interfaces · user modeling · device modeling · personalization · semantic web · WAI-ARIA · model-driven engineering · content adaptation · interface plasticity
Standards referenced: WAI-ARIA · CC/PP 2.0 · HTML5 · EARL