Alex Q. Chen

Web accessibility guidelines

The World Wide Web (Web) is in constant evolutionary change. This evolution occurs along many fronts and is led by infrastructure developers, Web designers, technologists, and users. These multiple stake–holders ensure that the Web is a heterogeneous entity, not just in the nature of the content, but in the technology and agents used to deliver and render that content. It is precisely this heterogeneity which gives the Web its strength and its weakness. A weakness in technology adoption leading to an increasing disconnect between the actual user experience and the expected experience of the technology stakeholders. We are interested in the human factors surrounding the evolution of the Web interface; and believe that the wait is always too long for new accessibility recommendations, guidelines, and technology to be adopted. In this case, we describe a ten-year longitudinal study comprising approximately 6,000 home pages. From this study we conclude that as a ‘rule-of-thumb’ mainstream technology is adopted at about 15% within the first three years, incremental version releases are adopted at about 10% within the first three years. However, sites which are most popular often exhibit enhanced adoption rates of between 10 and 15% over the same period. In addition, we see that accessibility guidelines are mostly ignored with only a 10% adoption rate after more than ten years. From this we infer that, for maximum accessibility adoption, guidelines might be supported and reflected in mainstream specifications instead of remaining only as a separate document.

The uptake of Web 2.0 technologies, and its impact on visually disabled users

World Wide Web (Web) documents, once delivered in a form that remained constant whilst viewed, are now often dynamic, with sections of a page able to change independently, either automatically or as a result of user interaction. In order to make these updates, and hence their host pages, accessible, it is necessary to detect when the update occurs and how it has changed the page, before determining how, when and what to present to the user. This can only be achieved with an understanding of both the technologies used to achieve dynamic updates and the human factors influencing how people use them. After proposing a user-centred classification of dynamic updates, this paper surveys the current state of technology from two perspectives: that of the developer, and those of visually disabled users. For the former group, the paper introduces some of the technologies that are currently available for implementing dynamic Web pages, before reporting on the results of experiments analysing current and historical Web pages to determine the extent of use of these technologies ‘in the wild’ and the trends in their uptake. The analysis shows that for the most popular 500 sites, JavaScript is used in 93%, Flash in 27% and about one-third (30%) use XMLHttpRequest, a technology used to generate dynamic updates. Uptake of XMLHttpRequest is approximately 2.3% per year across a random selection of 500 sites and is probably higher in the most popular sites. When examining dynamic updates from the perspective of visually disabled users, first an investigation is reported into which technologies (Web Browser and assistive technologies) are currently used by this group in the UK: Internet Explorer and JAWS are clear favourites. Then, the paper describes the results of an experiment, and supporting anecdotal evidence, which suggests that, at best, most users can currently reach updated content, but they must do so manually, and are rarely given any indication that any update has occurred. With technologies enabling dynamic updating of content currently deployed in about 30% of the most popular sites, and increasing annually, action is urgently required if visually disabled users are to be able to use the Web. The paper concludes by discussing some of the issues involved in making these updates accessible.

Widget Identification: A High-Level Approach to Accessibility

The Web 2.0 sees once static pages evolve into hybrid applications, and content that was previously simple, now becoming increasingly complicated due to the many updating components located throughout the page. While beneficial for some users, these components (widgets) are often complex and will lead to confusion and frustration for others, notably those for whom accessibility is already an issue. While users and developers often perceive widgets as complete components (a Slideshow, or an Auto Suggest List), they are in-fact heterogeneous collections of code, and are therefore hard to computationally identify. Identification is critical if we wish to reverse engineer inaccessible widgets or ‘inject’ missing ‘WAI-ARIA’ into ‘RIAs’. In this case, we introduce a technique that analyses the code associated with a Web page to identify widgets using combinations of code constructs which enable uniquely identification. We go on to technically evaluate our approach with the most difficult widgets to distinguish between—Slideshows and Carousels—and then describe two prototype applications for visually impaired and older users by means of example.

Widget identification and modification for web 2.0 access technologies (WIMWAT)

Rich Internet Applications (RIA) encourage World Wide Web (Web) content to be extracted and remixed from dif-ferent sources, so that presented content can be updated in small chunks, rather than reloading the entire Web page. These concepts change the way Web pages are created and how users interact with them. Hence, these changes will require assistive technologies to adapt to them. We intro-duce the concept of identifying and modifying embedded code within a Web page (widget) during development, so that widgets that are not in an accessible form, and which produce inaccessible content can be modi ed. Our concept can be also applied as an enhancement for screen readers, so that they can provide a preview facility of the types of features provided by the widgets. Currently, we are investi-gating di erent methods for detecting widgets from the Web pages source code. An evaluation of our detection methods has been carried out; this was an attempt to search for two types of widgets from the top twenty Websites. The evalu-ation successfully detected all widgets, suggesting that our identi cation methods are successful, although some false positives were also detected. Since we are in our initial stage, further investigations will be required to re ne our methods, and pursue the observation and modi cation phases required by the project.