Google DTorial: An interactive tutorial framework for blind users in a Web 2.0 world - Manual (Page 3)

accessible, with emphasis on improving Web accessibility [23, 26] for screen-reader
users. There has also been rapid development of tools to help Web designers check
for accessibility [14, 29]. Tools have also been developed to assist in the creation of
more accessible Web pages [5, 10]. While new screen-reader based solutions may be
invented, such solutions must be disseminated and adopted by the entire community
to be effective. However, if new technologies are created and implemented server-
side by Web developers, the burden of adopting new assistive technology will not be
placed on the millions of existing screen-reader users.
While prior research into help systems and tutorials is rich and provides a strong
foundation for future work, existing solutions do not take into account the unique
needs and challenges of visually-impaired individuals. Research into tutorials and
help systems [7, 19, 21, 22] has explored multimodal interfaces [16], behavior
modeling [32], and intelligent help systems [1, 12, 27]. Many solutions have been
created utilizing hypertext to provide easy access to help [11, 20]. Though some work
has been done on web based help systems [8, 28], it has largely been based in the
visual domain (images and Macromedia Flash [24]). Existing tutorials and help
systems use video, PDF, Flash, and HTML, while relying heavily on pictures and
animation. For VIUs, these modalities are inaccessible. The existing literature has not
suitably addressed methods for embedded tutorials in Web 2.0 applications or
targeted tutorial techniques for blind users.
The standard form of instruction for VIUs is a separate HTML page containing a
tutorial (and discovering the location of said tutorial is not always easy). Most HTML
tutorials are not screen-reader friendly, lacking appropriate HTML headings for easy
navigation and poor description of how to access/find content with a screen reader. In
addition, these tutorials lack context for the content and require frequent switching
between windows, often causing users to loose their “place” in both the tutorial and
application as they switch between the two windows.
With the advent of dynamic-asynchronous loading in 1999 [31], methods for users
to interact with Web pages drastically changed from the world of static HTML.
Dynamic Web pages, commonly referred to as Web 2.0 applications [15], often use
AJAX to enable content to dynamically change without a page reloading. As a result,
these Web applications function like a standard desktop application: the browser acts
as a platform on which these new applications run. This facilitates dynamic content
without page reloads and additional functionality such as application-based hot keys.
Consider a Web e-mail application where a user uses the “c” key to compose an email
or the “j” and “k” keys to cycle through the list of emails. Without JavaScript, this
functionality would only be available via mouse interactions and not via hot-keys.
Though AJAX provides useful features, Web 2.0 applications add complexity to
screen–reader users. Consider functionality in Appendix A, illustrating a JAWS
screen-reader interacting with Web pages. A typical screen-reader user operates in
two modes: Forms Mode (used for text input) and Virtual Cursor Mode (VCM) (for
page navigation and reading). Interaction is conceptually segregated into these two
modalities, making a natural division so users know when to switch modes.
Web 2.0 functionality (check marks in Appendix A) forces users to shift their
interaction with the Web application to a third mode, PC Cursor Mode (PCM), to use
Web-application specific keyboard commands. PCM stops JAWS from capturing
keystrokes and allows the user to invoke application-specific hot keys. However,