Click for better resolution

Project: Web Accessibility Scaffolding.

Concept: The concept of scaffolding, from the learning domain, allows users to learn certain skills. When applied to the accessibility domain, we argue that it helps users to function online with greater autonomy. In the case of visually cluttered Web pages, the scaffoldings expose the elements, pathways, and organization of a Web page that enable users to interpret and grasp the structure of a Web page. An accessibility scaffolding is more than just facilitation of task performance, it explicitly highlights and develops specific skills (i.e., finding core part of a Web page).

Three Examples: Interviews with advocates of users with diverse abilities, at the outset of this research, revealed that it is neither intuitive nor easy for the users to analyze and assimilate the overview for a page. We created three example scaffoldings; each representing a point in their respective design subspace that focus on certain accessibility skills to be acquired in dealing with cluttered Web pages.

• Position, speak, highlight first core text — teaches where core content begins
• Speak content using player interface — teaches how text flows in a Web page,
• Page-segment through button onterface — teaches what the overall structure of a Web page is.

Research: The concept of accessibility scaffolding was borne out of interviews with advocates of users with diverse abilities. In describing how they assist these users to work online, we learnt that they used coaching, teaching, and strategy sharing as techniques to guide and assist their users with ways to deal with visually cluttered pages. The three specific scaffoldings that were produced characterized specific difficulties experienced by the target user group.

Design: We exploited data on how the target user group used a tool (i.e., web adaptation tool) that had been developed for them earlier. Specifically the web adaptation tool consists of 12 sets of adaptations that include visual adaptations such as image enahancements, redundant presentations such as speak text, keyboard adjustment, and enlargement of browser control. Usage data of this tool revealed that all adaptations were used by some people. The single-most used adaptation (40% of the population) was ‘speak text’, wherein the user points the mouse at that portion of the page they want to have read aloud. Other adaptations, used by one third of this population, are font enlargement, font style changes, banner text and changing color contrasts.

Implementation: We made extensive use of a novel page-segmentation heuristic to realize the three scaffolding capabilities. This heuristic reconstructs the Web page layout and identifies the main regions and segments of a page using whitespace and regions as the primary criteria. It takes as input, the Document Object Model (DOM) tree information along with information about the graphical layout (i.e., position and size) of the DOM elements. The heuristic uses four passes to reconstruct the Web page’s layout. First, it finds the visual blocks of the Web page. Second, the visual blocks are sepa-rated into those belonging to the core region and border regions. Third, it determines whitespace separators between visual blocks in each of the core and peripheral regions. Finally, visual blocks are aggregated into larger units based on depth in containment hierarchy, separator size and visual properties of the visual blocks.

Evaluation: Full functioning versions of the three designs were shown to people working with seniors and people working with users with diverse development disabilities. Overall, our informants were uniformly positive about the first two scaffoldings (i.e., core content and speak text linearly in Web page with player interface). The seniors’ informants indicated that aside from the utility of taking the users explicitly to the start of the core content, this scaffolding bypasses difficult and problematic content that generally captivated and confused the seniors. The second group of informants recommended that the spoken text be shorter; on the order of a sentence construct.

Publication: Lee, 2004 and Lee and Hanson, 2003.

Work Conducted: November 2002 — November 2003.