Edapta. Device Independent Workshop Position Paper

Woman's Eye Woman's Ear Outstretched Hand

Edapta Position Paper

Device-Independent Workshop
3, 4 October 2000 -- Bristol, England

Kynn Bartlett, Director of Accessibility

Introduction to Edapta

Edapta is a San Diego-based startup founded in 1999. We focus on user-centric computing; specifically, on creating user interfaces which intelligently adapt to the needs or preferences of the user. We have a specific interest in the needs of aging users and users with disabilities, but our approach is applicable over the entire range of web access devices, from mobile computing to information appliances.

Our unique Edaptation Engine technology provides web sites with the intelligence needed to "morph" the user experience to meet user needs, representing the ultimate separation of content from presentation.

All users are not the same! Rather than relying on a single presentation (or a small handful) to meet the needs of anyone who would use a web application, our Edaptive technology creates a custom user interface optimized for each person. A blind web surfer need not hope for "graceful degradation" -- we will provide an interface optimized for linear access through a screenreader or Braille terminal. Access through WAP-enabled phones is no problem; WML is merely another dynamically generated user interface from our standpoint.

We apply the proven technology of "personalization" to the web user interface paradigm, operating not on the content level but on the level of the user's experience with the content. To date, the work in this field has been restricted to minor cosmetic changes such as changing the color of a heading or the font; we are revolutionizing web user interfaces by doing major structural and functional changes to the user interaction level in order to optimize the presentation and accessibility of the content.

Edapta relies heavily on the development and implemention of industry-wide non-proprietary standards, such as those developed by the World Wide Web Consortium. To be able to provide the ideal user interface, we will need a broad base of support for existing and emerging standards such as CC/PP, XHTML, and XSLT.

User Definition Objects and CC/PP

The W3C's Composite Capabilities/Preferences Profiles work enables web access methods to be self-identifying as to their own capabilities and their settings as functions of the user's preferences.

This self-identification is essential to the growth of the web as a user-centric communication medium accessible through a multitude of different access devices. The use of CC/PP profiles allows for server-side programs -- such as Edapta's Edaptation Engine -- to be written today which can understand the diversity of tomorrow's web access methods. CC/PP is a forward-looking method of dealing with variety of access methods which will blossom as the 21st Century unfolds, and is not limited to "how things look today", which has been a limitation on previous attempts to identify user agent functionality.

Edapta's user interface personalization technology uses User Definition Objects (UDOs) exchanged between various components of the integrated content delivery system -- these consist of the known qualities of the user and her system, including the CC/PP profile as well as any other information regarding the user's interaction with web content. Our UDOs accept and interoperate with CC/PP profiles.

Position: Edapta urges the W3C to continue to develop the CC/PP standards and looks forward to creating implementations which can interoperate with CC/PP-enabled user agents. We likewise hope that all browser and access device users will provide appropriate CC/PP profiles and identifiers to allow for a high level of interoperability.

Web Morphing and XSLT

Edapta's web morphing service operates under the direction of the Edaptation Engine (the "brains" of the operation) to create specialized presentation and interaction models for each user of an Edapta-enabled web site. A key to this process is Extensible Stylesheet Language Transformations (XSLT).

To provide optimized user experiences, we construct XSLT stylesheets on the fly and then apply them dynamically to the structured content to produce output in a presentational language (WML, XHTML + CSS, XSL-FO, VoiceXML etc.) appropriate for the user and her access method.

Position: Edapta is pleased to see continued development of specialized presentational languages based on XML (such as VoiceXML) as well as growing support for XSLT.

Content Storage and the Semantic Web

To produce an adaptive user experience, it is first necessary to separate the content from the presentation. This can be done in a number of ways; for example the use of XHTML 1.0 Strict plus Cascading Style Sheets attempts this separation.

In an ideal situation, the content would be stored in an XML-based, highly structured content definition language (or a database with an access method that duplicates such an XML application), without consideration of the ultimate presentation -- leaving that for the presentation layer. This content definition language is what HTML was intended to be. It is our position that XHTML does not currently fill this need adequately at the present time.

XHTML 1.0 (and 1.1) only provides the barest minimum in the way of structure for content, and is still reflective of the Web's origins as a distribution medium for a very specific type of document (physics papers). The content models in current XHTML do not reflect the use of the Web as it has evolved in use; XHTML structures have proven inadequate for describing the current applications of Web use. Furthermore, XHTML is intended to be used (with CSS) as the basis of a presentation layer directly; an ideal content definition language should not be directly used as the structure for creating the presentwation. The structure of the content and the structure of the display of the content are separate concepts, and current XHTML formulations do not take this into account.

A standardized approach to a content definition language which adequately reflects the structures of web documents would be very useful as it could be adopted by content management systems, web authoring software, and application servers -- employing a richer degree of semantics than currently available via XHTML. (The model of XHTML, however, could be used to produce modular content scheme based not on a vague sense of "what kind of markup should be used", but rather on the type and structure of the content to be presented.) XHTML could continue to function as a presentation layer structural language.

Position: Development of a standard language for presenting content structure in a way not currently available in XHTML will only help with the development of the Semantic Web and of tools and programs to process rich semantics to provide better adaptability to user needs.

Edaptive User Interfaces and the Web Accessibility Initiative

Edapta's progress in creating optimal user interfaces for users with disabilities or age-related changes has a solid foundation in the groundbreaking work of the W3C's Web Accessibility Initiative. Our approach is to serve as an example implementation of how dynamic content generation can be used proactively to meet the needs of people with disabilities and conform to the Web Content Accessibility Guidelines.

To date, most of the published guidelines from the WAI have (rightly) concentrated on the single source model of graceful degradation; until recently (with innovations such as CC/PP, XSLT, and XHTML), this has been the only reasonable approach to meeting users needs.

However, a single user interface cannot directly meet the needs of all users in the way that optimized user interfaces may; a solution which relies on "graceful degradation" is by definition a compromise solution. Especially when dealing with the needs of multiple disability groups, this has had the effect of introducing cases in which web designers and application developers have had to make difficult choices regarding who would and would not receive the best possible interface to web sites or applications.

An adaptive user experience can eliminate these problems and increase the usability (in addition to the accessibility) as each interface is constructed specifically for the person using it, and is not the result of an attempt to provide a single interface that can meet all user needs. For example, a "logical order" for visually displayed content (where position, size, and color are factors in how information is processed) is going to be very different from a logical order for linear users, such as those with screenreaders or Braille devices. By rebuilding the structure of the page for different access methods, we can provide an ideal experience rather than a compromise solution -- increasing user satisfaction (for Internet sites) and employee productivity (for Intranet/Extranet sites).

Position: The Web Accessibility Initiative should explore the issues of dynamically generated user interfaces and produce guidelines which cover situations in which a unique experience may be created for each user, as well as continuing to develop guidelines based on single user interfaces (as those will continue to be used for some time by many web authors).