Dynamic Accessible Web Content Roadmap

W3C Working Draft Draft September 23, 2005

This version:
http://www.w3.org/WAI/PF/roadmap/DHTMLRoadmap092305.html
Latest version:
http://www.w3.org/WAI/PF/roadmap/
Previous version:
http://www.w3.org/WAI/PF/Group/roadmap/DHTMLRoadmap082905.html
Editor:
Richard Schwerdtfeger, IBM <schwer@us.ibm.com>

Abstract

W3C roadmap for addressing the accessibility of dynamic web content.

Request for Comments

The PF working group has developed this roadmap for making dynamic web content more accessible to people with disabilities. To assure this roadmap and the corresponding technologies produced to support it will interoperate in a manner that improves accessibility for the widest range of web technologies, the PF working group asks for comments and suggestions on how this roadmap can be improved for use in the technologies that you or your working group are developing. In general we prefer public comments so that all people (including the general public) reviewing the document can be aware of your comments, but you may also send member confidential comments that will be available only to the working group and other W3C members.

Status of this Document

This is the latest draft by Rich Schwerdtfeger of the Protocols and Formats working group.

Table of Contents

  1. Dynamic Web Content Roadmap
  2. 1.0 The Problem
  3. 2.0 Requirements
  4. 3.0 Proposed Solution
    1. 3.1 Executive summary
    2. 3.2 HTML Accessibility API Support Analysis
  5. 4.0 The Roadmap - Filling the Gaps
    1. 4.1 Use of new provisions for keyboard focus and semantics to support platform accessibility APIs
      1. 4.1.1 Provision for the Role attribute "What the object is"
      2. 4.1.2 Provision of the accessibility state information "What meaningful properties does this object have at this time?"
      3. 4.1.3  Provision of the keyboard or input focus "What object am I working on?"
    2. 4.2 Use of XHTML2 Role landmarks to improve navigation
    3. 4.3 WAI Role Taxonomy for Accessible Adaptable Applications - Define XML Semantics for role through RDF
      1. 4.3.1 Future Opportunities for Role and Semantic Web technology
    4. 4.4 Accessibility Events and Event Handling
    5. 4.5 XML Events, XForms, and XHTML2, and declarative markup - A Look Ahead
      1. 4.5.1 XML Events and Handlers - "adding purpose to actions"
      2. 4.5.2 XForms
      3. 4.5.3 XHTML2
    6. 4.6 User Agent Support
    7. 4.7 Assistive Technology Support
    8. 4.8 Authoring Tool Support
    9. 4.9 Developer Guidelines and Implementations
    10. 4.10 Roadmap Deliverable Timeline
  6. 5.0 Conclusions
  7. 6.0 Appendix

Dynamic Accessible Web Content Roadmap

JavaScript is found on over 50% of all web sites today, dramatically affecting the ability for persons with disabilities to access web content. New Rich Internet web applications are being created to render custom widgets modeling rich desktop componentry. Legacy GUI accessibility frameworks address these issues through a comprehensive accessibility API and infrastructure to foster interoperability with assistive technologies. These API's constitute a contract between applications and assistive technologies, such as screen readers, to allow them to gain access to rich dynamic content with the appropriate semantics needed to produce a usable alternative.

Unfortunately, HTML and other markup does not provide adequate markup to support accessible dynamic content. Until now, the W3C WAI has discouraged the use of JavaScript per Web Content Accessibility Guidelines (WCAG) 1.0. The W3C has a number of initiatives in the pipeline to address this problem using a declarative markup approach. This roadmap is designed to create a bridge to fix the interoperability probem with assistive technologies today by incorporating the appropriate meta data in today's XHTML markup to support today's accessibility API. It will incorporate some of the advanced accessibility features in technologies like XHTML2. The intent of XHTML 2 is to make the enablement of web applications and documents easier for authors. This roadmap will create cross-cutting technologies which may be incorporated by XHTML 2 authors in their markup to describe GUI widgets and document structure to assist assistive technology vendors in providing accessible, usable solutions. The W3C WAI PF working group will work with User Agent manufacturers and assistive technology vendors to ensure we have a working solution.

1.0 The Problem

Accessibility for traditional HTML suffers from a number of problems that make it difficult to support dynamic content:

  1. Accessibility is dependent on abstracting semantics from both content and presentation information. Extracting this semantics from today's HTML content is typically not reliable. Today, semantics are limited to the tag elements names.
  2. HTML allows content to be repurposed for presentation formatting without providing a way to convey this information. A common example of this is the use of tables for formatting content as opposed to style sheets.
  3. HTML, when combined with script and CSS may be repurposed to create dynamic custom components without the ability for the author to convey this information to the native accessibility architectures designed to support dynamic GUI content.
  4. HTML lacks the ability to attach meaningful meta data about document structure.

Authors of JavaScript generated content do not want to tie themselves to the standard tag elements that define the actual user interface element such as tables, ordered lists, etc. JavaScript also make extensive use of tag elements such as DIVs where they will dynamically apply a UI through the use of style sheets and dynamic content changes. An HTML DIV provides no semantic information. The page author may define a DIV as the start of a pop-up menu or even an ordered list. No HTML facility provides a mechanism for the page author cannot convey the role of the DIV as being a pop-up menu. Consequently, a page author cannot tell an assistive technology when these elements have focus. There is no mechanism to convey accessibility property information as to whether the pop-up menu is collapsed or expanded. There is also no mechanism to define what the actions are that can be formed on the element other than through a device-dependent means through the event handler type (onmouseover, onclick, etc.)

In short JavaScript needs an accessibility architecture it can write to such that a solution could be mapped to the accessibility frameworks on the native platform by the user agent.

Figures1.0 Accessibility Interoperability at a DOM Node without JavaScript

Accessibility infomation mapped to a DOM element in the DOMt

Figure 1.0 illustrates a typical DOM node in a Model-View-Controller architecture. On the node, data, or the "Model", which should include semantic information, is separated from the user interface presentation, the "View." Here, the document element is managed by the user agent based on the default behavior of the element. The user agent's default behavior at the document element forms the controller. Placed between the DOM node and the asssistive technology is a box containing the contract provided by the user agent to the assistive technology. This data includes typical accessibility information found in the accessibility API for many of our accessible platforms for GUIs (role, state, caret, selection, text, hypertext, name description,  parent/child information, parent/child information, and relations). For HTML and other W3C markup, the accessibility information provided is dependent on that which is provided by the element's tag name and any accessibilility attributes that map to that provided. For example, the accessible role of a table is table. The accessible description would be provided by the author by assigning a title attribute.

Figure 2.0 Accessibility Interoperability at a DOM node with JavaScript)
DOM Element with JavaScript controller

Figure 2.0 shows the same DOM node provided in Figure 1.0 but with JavaScript acting as the new controller. JavaScript overrides the default user agent behavior at the DOM node. It manipulates the data, content, and style in response to events caused by user interaction to produce Custom widgets. In this situation the default accessibility information is no longer valid and therefore the contract is now invalid. Figure 2.0 shows the contract with asterisks in front of role, state, actions, value, event changes, and relations. These asterisks represent potential accessibility errors and gaps in the base markup. These gaps result from the author's inability to provide the new semantic data needed to support the contract.

2.0 Requirements

Allow for discovery of custom UI components through the use of semantic web technologies

Web ontologies allow for storage of semantic information about objects and how they relate to others in the ontology.

Solution must support today's accessibility architectures

Accessibilty architecture today is centered around object technology. Each object in an application or document, exposes accessibility information about itself to an assistive technology.

Allow for separation of content and presentation

Dynamic content authors must be able to store the accessible meta data in the document independent of how it is rendered.

Allow for passive monitoring of the application by an assistive technology

Assistive technology vendors should not be required to poll an application for changes in accessibility information.

Leverage new W3C efforts to solve the problem

This problem needs to be solved quickly. There are a number of efforts underway, such that minimal changes may be required to bring them to the level needed.

Light weight

The solution needs to be light-weight in order to see uptake among HTML authors.

Scaleable

The solution needs to be scalable, i.e., make simple things easy while making complex things possible.

Internationalizable

Like other Web solutions, our solutions must be internationalizable.

Guarantee User Agent support up front

We need to involve user agent manufacturers up front

Involve assistive technology vendors up front

Assistive technology vendors need to be involved from day one. We need to leverage support by AT vendors so that we can have a total solution available when the specification is complete

Produce Developer Guidelines during the process

This is a critical mistake made by people creating a new accessibility model. Developers must be on board early on so that they may contribute feedback and start producing workable solutions early.

3.0 Proposed Solution

3.1 Executive summary

What is clear from the problem statement is that the author does not have the ability to provide the appropriate accessibility information in the markup to support the accessibility APIs on the target platform. This problem is not limited to HTML. It extends to other markup, including SVG. Some of these issues will be addressed through the introduction of declarative markup. New work being done on XForms and XHTML 2 are designed to incorporate accessibility features through the declarative markup of the language. This has the added benefit of reducing the enablement effort by authors through leveraging the existing accessibility information stored in these markups to offload some of the accessibility work to the User Agent. That said, we need to build a bridge which will fill the accessibility gaps on today's HTML markup which will lead to broad applicability to today's markup while moving forward with declarative markup to reduce the enablement effort.

3.2 HTML Accessibility API Support Analysis

Using Figure 1.0 as template for addressing the problem and 508 accessibility standards, a Gap analysis has been constructed in Table 1.0 to show where the gaps are in the infrastructure and what W3C standards should be used to address the problem. In right column, table cells which are empty or which indicate a limitation represent accessibility gaps in HTML and XHTML.

Table 1.0 Platform Gap Analysis for Accessible Dynamic Web Content for HTML and XHTML

Required Components Who does what today?
(HTML)
Events:  
FocusChange DOM 2, 3 events
Activation
User Agent API
Caret Change User Agent API
Value Change  
State Change  
Selection Change User Agent API
Mutation DOM Events
Accessible Actions:

Event Handler functional information to label the actions
 
Access to the available event handlers for enumerating the actions

State Information:
Role Information: Limited to standard HTML tag names. (Mix Content/presentation)
Relationships: Parent/child Limited DOM (affected by style) (Mix Content/presentation)
Relationships: (Label, Group, ControllerFor) Limited to HTML (Title, alt, label)
Text Core DOM from parsed HTML
Content selection: browser dependent (incomplete)
Font/Font Style Information: Can set but can’t get final format
Description/Help: limited to HTML 4.0 - Alt Text, title text
Accessible value: limited to form elements
Coordinates (Bounding rectangle, etc.): User Agents.platform accessibility API
Accessible Name:  
Respond Desktop Font/Color Changes: partial (conflicts with CSS and JavaScript)
Device independent navigation:
Accessibility API Mapping: partial - User Agents
Provide focus to all active elements
(important for equivalent keyboard access on desktops)
Limited to forms and achors


4.0 The Roadmap - Filling the Gaps

At this time, the primary effort in the W3C WAI Protocols and Formats working group is to focus on an extension to XHTML 1.1. It should be noted that XHTML 1.0 could also be enhance to support this extension through the use of a hybrid DTD which incorporates the extensions. The reason for focusing on XHTML 1.X is that XHTML is extensible through the use of namespaces and because it is renderable by today's browses. For this work, we will be producing a document called the Accessible Adaptive Applications (AAA) document and a GUI role taxonomy document. The AAA will constitue extensions to XHTML 1.1 to fill most of the gaps needed to support accessibility API infrastructures and dynamic XHTML content. It will also borrow the role attribute from XHTML 2 for which a set of accessible role values may be applied from the WAI GUI Role Taxonomy specification. The Role taxonomy specification will provide the necessary core roles found in Accesssibility API sets for Windows and Linux. Additionally, it will contain roles which are representative of document structure. The use of document structure is necessary for assistive technologies to navigate complex documents and to know when they have entered active areas of a web page such as in the case of a dynamic scripted web application. It will be designed to assist the user agent or authoring tool in determining what properties are supported by a given role and assist with accessibility api mapping of these properties. The taxonomy will be like a class hierarchy used to convey semantics and structure and therefor knowledge about each role. At this time that taxonomy will be written in RDF.

The GUI Role Taxonomy specification is meant to be cross-cutting. XHTML 2 introduces the role attribute as part of its built-in accessibility support. The common roles defined in this taxonomy may also be applied to XHTML 2 content and other markup such as SVG should those efforts allow for a mechanism to reference these common roles.

Table 2.0 GAP Analysis - Progressive Filling of the Gaps for HTML in support of Accessibility API

Required Components Who does what today?
(HTML)
Plan to fill the gaps for XHTML 1.X
Plan to fill gaps for
XHTML 1.X + XForms
XML Events Plan to fill the gaps for
XHTML 2.0
Events:  



FocusChange DOM 2 UI Events



Activation
DOM Level 2 Device Independent UI Events



Caret Change User Agent API User Agent API User Agent API

User Agent API
* Value Change   User Agent to monitor specific accessibility properties in the AAA. For properties representing platform accessibility API values as specified by a given role, it will generate an accessible value change event. (See Value Change)


User Agent to monitor change from WAI State Taxonomy information  supplied by the Author
and fire the platform specific Accessibility API event
* State Change   User Agent to monitor specific accessibility properties in the AAA. For properties representing platform accessibility API states, it will generate an AccessibleState Change event. (See State Change Event)

User Agent to monitor change from WAI State Taxonomy information  supplied by the Author
and fire the platform specific Accessibility API event
Selection Change User Agent API
User Agent API
User Agent API

User Agent API
Mutation DOM Events



Accessible Actions:




* Event Handler functional information and descriptive shortcus
 

  • XML Event handler "purpose" addition to XML Events

Details:Need for XML Events
  • access element
  • XML Event handler "purpose" addition to XML Events

Details:Need for XML Events*
* Access to the available event handlers for enumerating the actions DOM 3 Events




* State Information:   Supplied by the AAA specification. User Agent to  map information found in XForms data model
with limited use of  AAA.

User Agent of XForms data model to accessible state meta found in:

(WAI States and Adaptable Properties Module)

User Agent to  map information found in XForms data model
with limited use of  AAA.

User Agent of XForms data model to accessible state meta found in:


(WAI States and Adaptable Properties Module)

* Role Information: Limited to standard HTML tag names. (Mix Content/presentation)


Relationships: Parent/child Limited DOM (affected by style) (Mix Content/presentation)



Relationships: (Label, Group, ControllerFor) Limited to HTML (Title, alt, label) AAA Specification

  • AAA Specification
  • label, group acquired from XForms


label, group acquired from XForms incorporated into XHTML 2
Text: Core DOM 1 from parsed HTML



Content selection: User Agent




Font/Font Style Information: Can set but can’t get final format User Agents final format styling using CSS2 APIs
User Agents final format styling using CSS2 APIs
User Agents final format styling using CSS2 APIs
* Description/Help: limited to HTML 4.0 - Alt Text, title text AAA Specification
Acquired from XForms help, hint
Acquired from XForms help, hint incorporated into XHTML 2
* Accessible value: limited to form elements
AAA Specification
User Agent acquires from XForms data model
User Agent Acquires from XForms data model incorporated into XHTML 2
Coordinates (Bounding rectangle, etc.): User Agents.platform accessibility API




Accessible Name: User Agent determines from HTML tag




* Respond Desktop Font/Color Changes partial (conflicts with CSS and JavaScript) CSS3 - Requires User Agent support
CSS3 - Requires User Agent support
CSS3 - Requires User Agent Support
* Device independent navigation: Use of xhtml2 landmarks through the <link rel"">
XHTML 2 Role land marks

XHTML 2 Role land marks
XHTML 2 Role land marks,  and Access Key replacement
*Accessibility API Mapping: partial - User Agents
User Agent to map new AAA properties and role meta data to the platform accessibility API



Provide focus to all active elements
TABINDEX=-1 best practice
TABINDEX
TABINDEX

All XHTML 2 elements shall be focusable.

Table 2.0 shows the roadmap for filling the gaps in Table 1.0. The effort is designed to fill the gaps for HTML content. Required components marked with an asterisk indicate gaps to fill in the roadmap. As you move to the right in the table a transition to more declarative markup will take place. The use of declarative markup reduces the enablement effort required while better addressing accessibility. Declarative markup is used to build accessibility in rather than bolting it on. Subsequent sections highlight the requirements and standards efforts set to address the gaps.

A way to summarize this chart is that the critical extensions needed to make accessible dynamic web content accessible for XHTML 1.x, that may be made interoperable with assistive technologies, will be provided through the use of:

  • AAA Specification
  • XHTML 2 role attribute
  • XHTML 2 role document landmarks
  • WAI GUI Role Taxonomy Specification
In short, these changes will fix the dynamic acripted web content, in particular the use of JavaScript with HTML markup.  The next section will describe how these are used all together. Less critical for XHTML 1.X but will be very helpful for accessibility will be the descriptive extensions to XML events and the new access element for XHTML 2.

4.1 Use of new provisions for keyboard focus and semantics to support platform accessibility APIs

Adaptive technologies, that need to provide alternative access to complex user interfaces authored via HTML, are often left guessing at the semantics behind specific portions of HTML document. As an example, an XHTML document might use a certain HTML construct, such as a collection of DIVs, to create navigation bars, a site-navigation menu, and other GUI-like user interface widgets. To fix the problem we will cover how we incorporate the role attribute, the accessible state properties, and give the object focus using the new TABINDEX feature. Addition of this information will allow the author to provide the necessary information to enable the user agent to support the accessibility api accessed by the adaptive technology.

4.1.1 Provision of the role attribute "What the object is"

Each platform accessibility API has the notion of a role for a GUI object. This is the case for Microsoft Active Accessibility (MSAA), the Java Accessibility API, the Gnome Accessibility Toolkit (ATK), the Mac Accessibility API for Cocoa, or UI Automation for Longhorn (called content type in UI Automation). To add this information we use namespaces to pull in the role attribute from XHTML2 and assign it a value from the GUI Role Taxonomy. Since the XHTML 2 role attribute takes a qname we can use a qname to reference the role attribute from the W3C WAI GUI RoleTaxonomy. In the following example we use namespaces to assign a role of menu to a div. Menu is defined in the GUI role taxonomy.

Example: Use of namespaces to incorporate role information information into XHTML 1.x
    <?xml version="1.1" encoding="us-ascii"?>
<!DOCTYPE html PUBLIC "Accessible Adapative Applicaitons//EN" http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml"
xmlns:x2="http://www.w3.org/2002/06/xhtml2"
xmlns:wairole="http://www.w3.org/2005/01/wai-rdf/GUIRoleTaxonomy#"
>
<body>
<div x2:role="wairole:menu">
File
</div>
</body>
</html>

A decision by the HTML working group was to make role a qname and allow others to define sets of standard roles. Furthermore, these roles should be extensible. It is the intent of the HTML working group to have these qnames refer to RDF which defines the element. The RDF should ultimately define the object. The WAI Role Taxonomy for Accessible Adaptable Applications, covered later in this roadmap, introduces the purpose of this taxonomy and how RDF can be used to describe these roles.

4.1.2 Provision of the accessibility state information "What meaningful properties does this object have at this time"

Since this is dynamic content, the state of these new repurposed objects will change. The WAI Accessible States and Adaptable Properties Module shall provide the common accessible properties needed to support the accessible state or property information provided by the platform accessibility API defined previously. This specification was created based on an analysis of the accessibility properties defined in MSAA and ATK. To pull in these information we again use namepaces to extend XHTML. The following example extends the previous by adding the haspopup accessibility property.


Example: Use of namespaces to incorporate accessible state information information into XHTML 1.x
    <?xml version="1.1" encoding="us-ascii"?>
<!DOCTYPE html PUBLIC "Accessible Adapative Applicaitons//EN" http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml"
xmlns:x2="http://www.w3.org/2002/06/xhtml2"
xmlns:wairole="http://www.w3.org/2005/01/wai-rdf/GUIRoleTaxonomy#"
xmlns:waistate="http://www.w3.org/2005/07/aaa"
>
<body>
<div x2:role="wairole:menu" waistate:haspopup="true">
File
</div>
</body>
</html>

A Windows user agent may now map this property to the Microsoft Active Accessibility state of STATE_SYSTEM_HASPOPUP. Adding or removing this state would result in the Windows user agent sending an EVENT_OBJECT_STATECHANGE event to the assistive technology. The task of the JavaScript page author would be to maintain this state attribute which can easily be done through the use of Document Object Model calls.

4.1.3 Provision of the keyboard or input focus "What object am I working on?

Virtually all adaptive technology solutions, such as screen readers and onscreen keyboards, need to know what object currently has focus. For example, I might want to insert text into the current object with focus or I might want to announce information about the object that has focus. Today, HTML 4.01 and XHTML 1.x limit script authors to only being able to provide focus to form and anchor elements yet the Document Object Model Specification allows all elements to receive events including keyboard events. This means that HTML, by design prohibits script authors from making all HTML elements keyboard accessible. This single problem has impacted the usability of web pages whereas gaining access to all elemens means using the tab key on desktop browsers. This slow, unproductive, approach makes it difficult for portal navigation whereas all active elements must be tabbed through to get to an active element in the last portlet in a document.

To solve this problem in XHTML 1.x we are incorporating a feature in Firefox and IE to define the tabindex for -1. This allows a script author to give an element focus without placing it in the tab order: The following table desribes these changes which will be incorporated into the new Accessible Adapative Application specfication.


tabindex attribute Focusable with mouse or JavaScript via element.focus() Tab navigable
not present Follows default behavior of element (yes for form controls, links, etc.) Follows default behavior of element
Negative, e.g. tabindex="-1" Yes No, author must focus it with element.focus() as a result of arrow or other key press
Zero, e.g. tabindex="0" Yes In tab order relative to element's position in document
Positive, e.g. tabindex="33" Yes Tabindex value directly specifies where this element is positioned in the tab order.
These elements will be positioned in the tab order before elements that have tabindex="0" or that are naturally included in the tab order (form elements and links)

The following example shows the introduction of TABINDEX to provide focus to a DIV having the new accessibility meta data:

Example: Use of tabindex to give non-form and anchor elements focus in XHTML 1.x
    <?xml version="1.1" encoding="us-ascii"?>
<!DOCTYPE html PUBLIC "Accessible Adapative Applicaitons//EN" http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml"
xmlns:x2="http://www.w3.org/2002/06/xhtml2"
xmlns:wairole="http://www.w3.org/2005/01/wai-rdf/GUIRoleTaxonomy#"
xmlns:waistate="http://www.w3.org/2005/07/aaa"
>
<body>
<div x2:role="wairole:menu" waistate:haspopup="true" tabindex=-1>
File
</div>
</body>
</html>


XHTML 2 will allow all elements to receive focus, eliminating the need for this tabindex modification.

4.2 Use of XHTML2 role landmarks to improve document navigation

In addition to the common roles which will reside in the GUI Role Taxonomy, XHTML 2.0 defines a collection of common roles that define pertinent parts of a document for the purpose of accessibility. User agents may incorporate device equivalents, such as key mappings in the case of a desktop user agent, to navigate to these sections of a document independent of the web site. The addition of these semantics allows the user agent to provide standardized navigation to common document sections. This is especially important for portals to improve the usability. These may be used as attributes in XHTML 1.x by applying them to sections of the document as shown in this example.

Example: Use of namespacs to incorporate an XHTML 2.0 common landmark to define a landmark for the navigation section in an XHTML 1.X document
    <?xml version="1.1" encoding="us-ascii"?>
<!DOCTYPE html PUBLIC "Accessible Adapative Applicaitons//EN" http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<html xmlns="http://www.w3.org/1999/xhtml"
xmlns:x2="http://www.w3.org/2002/06/xhtml2"
xmlns:wairole="http://www.w3.org/2005/01/wai-rdf/GUIRoleTaxonomy#"
>
<body>
<div role="x2:navigation">
<ul>
<li> <a href="www.w3.org/WAI">WAI</li>
</li>
<li>
<a href="www.ibm.com/able">IBM Accessibility Center</a>
</li>
<li>
<a href="www.opera.com">Opera Software</a>
</li>
<li>
<a href="www.mozilla.org/access">Mozilla Accessibility</a>
</li>
<li>
<a href="www.ubaccess.com">UB Access</a>
</li>
</ul>
</div>
</body>
</html>


An an alternative to placing these landmarks on the XHTML elements is to place them in the header through the use of link which is also applicable to HTML 4.01. Here, Link's rel attribute may take an string value which may be used to represent the landmark. Additionally, a description may be added to the title as shown in this example:


Example: Use of Link/Rel  with an ID href and title to assign a document landmark

<html lang="en-us">

<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" >
<title>Career Center Home Page</title><link rel="x2:banner" href="#title" title="Career Center">
<link rel="x2:navigation" href="#nav1-links" title="Career Counseling Resources">
<link rel="x2:navigation" href="#nav2-links" title="Resources by Audience">
<link rel="x2:navigation" href="#leftnav-links" title="Career Center Services">
<link rel="x2:navigation" href="#bottomnav-links" title="Quick Links">
<link rel="x2:main" href="#content-col-1" title="Welcome to the Career Center">
<link rel="x2:secondary" href="#content-col-2" title="Featured Events">
<link rel="x2:secondary" href="#contact-info" title="Contact Info">
<link rel="x2:search" href="#search" title="Search Career Center website">
<link href="styles/career-main.css" type="text/css" rel="stylesheet">
</head>

This is being worked by the WAI coordination group as a best practice. The example above was taken from the header from the Career Center web page at the University of Illinois at Urbana-Champaign.

Students form this university, under the Jon Gunderson's guidance created Accessibility extensions for Mozilla/Firefox, in part, allow a page author or user to launch a table of contents to viewer to list the landmarks. Keyboard navigation of the table of contents causes the corresponding document section to be highlighted.

Figure 3.0 Table of Contents generated from navigation landmarks in the header

Table of Contents from Landmarks

Figure 3.0 shows the Accessibility extensions for Mozilla/Firefox from the University of Illinois at Urbana-Champaign to render the document landmarks. This picture shows the Firefox Browser rendering the University of Illinois Career Career Center home page. In this example. The "Table of Contents" viewer is shown launched from the extension on the toolbar and the secondary "Featured Events" is selected resulting in that section of the document to be highlighted in yellow. The Table of Contents viewer lists the following section type/title pairs:
  • Banner / Career Center
  • Navigation / Career Counseling Resources
  • Navigation / Quick Links
  • Navigation / Resources by Audience
  • Main / Welcome to the Career Center
  • Secondary / Content Info
  • Secondary / Featured Events
  • Search / Search Career Center

4.3 WAI Role TAXONOMY for Accessible Adaptable Applications - Define XML Extensible Semantics for Role through RDF

It will be necessary to describe how to interact with a document element having a specific role. This is critical when defining roles not included in the common set specified previously. It should be possible to describe UI widgets and how to interact with them. RDF is a way for using data to describe data and provides a W3C standards-based approach to represent this information.

Figure 4.0 Partial RDF Map for a GUI Role Taxonomy

Sample Semantic Map for Taxonomy 

Figure 4.0 Shows a basic RDF mapping which defines a set of terms and relationships defining an object. At the center is a Widget object that defines common states and properties for all GUI widgets. The Button object extends Widget and inherits defined accessibility properties from the superclass Widget. It also defines a relatedConcept property to a Link object. The ButtonUndo role extends Button and an instance of extends of Button. It  has a relatedConcept of an HTML input object. ButtonUndo will introduce Dublin Core meta data such as the description of the object.

The terms relatedConcept and requiredState are terms which which will be defined as part of the corresponding RDF schema. Each role instance will represent standard Roles found in platform accessibility APIs platforms like Windows and Gnome as well as content structure. These roles will form the taxonomy. Each RDF role may be referenced  using a qname from Host XML markup such as XHTML:


<div x2:role="wairole:ButtonUndo">

whereby wairole:ButtonUndo expands to:

http://www.w3.org/2005/01/wai-rdf/GUIRoleTaxonomy#ButtonUndo in the button markup

The power of this design is it allows user agents to go back into the corresponding RDF markup and determine what properties it supports for Accessibility API mapping. Also, new web authoring tools can use the self describing RDF to help test for accessibility conformance to semantic web markup representing the role. Additional, middleware solutions can now make intelligent transformations of Web Content by processing the semantics behind rich browser and rich structured frameworks to adapt accessible solutions for a broader user base.

Our immediate goal is to fix the accessibility problem with scripted web content.

Assistive technologies will use the standard roles to determine how to render most content. For custom widgets they should be able to determine how to interact with it using the RDF document defining the roles.

Figure 5.0 DHTML Example

DHTML example of GUI-like notebook tab with a data drid

Figure 5.0 shows a DHTML example of using XHTML, JavaScript, and CSS to produce a GUI-like application. This example of a DHTML web page being developed in IBM shows a notebook tab with a data grid that behaves like you would expect a GUI to. Keyboard navigation within the data grid is done with arrow keys. Navigation among the page tabs is done with the arrow key. Navigation between the notebook tab, the edit fields, buttons, and the data grid is done using the tab key. Accessible role and state meta data from the GUI role taxonomy and Accessible Adaptable Application specifications, respectively are added as attributes to each of the XHTML elements repurposed as GUI widgets dynamically. The user agent, in this case Firefox, maps this information to the platform accessibility API. Figure 6.0 shows the Microsoft Active Accessibility rendering of the new accessibility markup provided on the DataGrid page tab which has focus.

Figure 6.0 Microsoft Inspect Tool rendering of the page tab DataGrid

MSAA Inspect Tool diagnostics for Notebook page tab

Figure 6.0 is a Microsoft Inspect 32 rendering of the DataGrid Page page tab in Figure 5.0. Inspect32 provides Microsoft Active Accessibility Information. It shows the accessible role of "page tab" and accessible state information of focused, focusable, and linked. There are no page tab elements in XHTML. Here, an  XHTML DIV element is repurposed by a JavaScript controller to look like a notebook tab. It is now able to receive focus unlike in standard XHTML 1.X. It can do so without requiring tabbing. With  these specifications the script author may now add the accessibility properties to support platform accessibility API. Accessible state properties for the DataGrid page tab are shown as focused, focusable, and linked. Unlike a GUI application, the author need only enable their application once for multiple operating system platforms.

4.3.1 Future Opportunities of Roles and Semantic Web Technology

Beyond scripted web content, the working group intends to extend the use of roles to enable other user cases. These may include:

  • Structured Textual Markup - enhancing structure of the markup of a document, including Data Tables , or translating the structure of an XML document to a markup structure that user agents are used to dealing with (e.g. myXML to XHTML) Binding sections of a document to a common role. This allows for different navigation techniques though a document
  • Knowledge representation of web content - As a secondary benefit, Roles improves compatibility Knowledge-Based Services and the Semantic Web. In integrating accessibility and the semantic web, accessibility can be moved forward, paving the way for customized accessible searches and intelligent user agents with additional applications.
  • Adding concepts in the type of content for adaptation to the user scenario - The more that is understood about content the better it can be adapted for the end user. For example:

    1. If it is known that a page hyperlink has the role of taking the user to the site’s home page, then that knowledge can be used to create enhanced accessibility in different ways in many different scenarios, such as icons or accesskeys.
    2. If it is known that a text box is for the user email address then the user agent can support users filling in the form by labeling it with an icon or symbol, automatically validating it, or even form filling.
    3. If it is known that a paragraph is complex, a simple equivalent can be shown in its place
    4. If a word is ambiguous then a role of a concept can be give, providing clarity
       An example of this may be :

      <span x2:role="role:nonliteral" aaa:hasAlternate="no">

4.4 Accessibility Events and Event Handling

Interoperability between applications and assistive technologies requires event notification for accessibility. The events defined in Table 2.0 will be fired via the user agent. The accessible value and state property changes will be generated in response to changes in the DOM attributes as defined by the AAA specification. User agents supporting the platform accessibility API, will support event notification such as the MSAA EVENT_OBJECT_STATECHANGE or EVENT_OBJECT_VALUECHANGE events.

4.5 XML events, XForms, XHTML2, and declarative markup - A Look Ahead

Up to this point, the roadmap has discussed addressing today's XHTML markup as supported by today's desktop browsers. The markup in today's browsers, like accessibility, requires developers to do things manually through the use of JavaScript. While this is extremely flexible it suffers from a number of problems:
  • Although improvements have occurred, JavaScript is not the same on all browsers and platforms
  • Data models are inconsistent
  • Accessibility is bolted on rather than being built in
  • Accessibility solutions are more device dependent than we like
  • Heavy accessibility lifting is done by the author in line with the way JavaScript is used
Industry evolution has shown that we can do things better and produce more consistent cross-platform solutions which support accessibility and reduce the enablement efforts of the author.

4.5.1 XML Events and Handlers - adding purpose to actions

HTML is a device dependent solution. An impaired user needs to make an intelligent decision as to what actions they can perform on a document element. An impaired user wishing to enumerate the available actions that may be performed on a page element is confronted with the non-informative information provided by HTML's intrinsic events. The information provided is very device specific and non-informative. For example, what action does an event handler perform in response to an onclick event? This is problematic for blind, low vision, and mobility impaired users. There may be more than one action to be performed on a given element. A mobility impaired user would like to enumerate the handlers and make a decison. Generating the appropriate mouse event is problematic. Blind users and low vision users would like to make informed decisions before they accidentally make a purchase. Some information can be gleaned from the UI by sited users, however some of this information may be obscured from view due to the use of a magnifier by low vision users.

The HTML working group is working to address this problem through the use of XML events. XML Events provide XML languages with the ability to uniformly integrate event listeners and associated event handlers with Document Object Model (DOM) Level 2 event interfaces [DOM2EVENTS]. The result is to provide an interoperable way of associating behaviors with document-level markup. Through the use of XML this is done declaratively. Having the ability to use XML to integrate listeners and handlers will allow us in in future versions of the XML event specification to tie a descriptive purpose to the handlers.

This problem also manifests itself when encountering event handlers authored via element script. XML Events and XML Handlers solves the first half of the accessibility problem by introducing a clean declarative syntax while encouraging authors to use high-level event names and descriptions. This gives the author the flexibility to assign device specfic mappings to event handlers while providing for added semantics. In XForms 1.0, we have gone one step further by introducing declarative action handlers for commonly used event handlers, e.g., setfocus. The combination of REL navigation semantics, XML events, XML handlers, and XForms declarative markup set the stage for an access key replacement. Access key suffers from a lack of semantics and support for device independence.

Dynamic web content implies the use of event handlers. From an accessibility perspective we need to assist the user in determining how to interact with a web application in a device independent way. The provision for named events and the ability to bind event handlers to document element events in the  XML Event specification is critical. At the time of binding we can incoporate the additional Role information as well as make use of the name information to provide semantics about what the event handler does. In the future, this information should be mapped to the available accessibility infrastructure by a, UAAG conforming, user agent on the platform. Once complete, a user agent conforming to the  DOM 3 event specification will allow for enumeration of event handlers with corresponding descriptions. The ability to ask this information has been added to the EventTarget interface in DOM3 events through the hasEventListenerNS method.

Platform accessibility API like the Gnome ATSPI Accessible Action interface may now enumerate these actions for assistive technologies so that the user may be provided with descriptive information to make an informed decision as to which action to take.

Screen Reader user would be able to pop-up a list of event handlers in a list such as follows:

  • Role over help information
  • Submit the form
vs:
  • OnMouseOver
  • OnClick

Authors wishing to incorporate XML events may do so to existing XHTML 1.1. markup through modularization.

4.5.2 XForms

XForms addresses the accessibility problem in the following ways:

  1. Separates content from presentation
  2. Doing the first has resulted in semantic meta data stored in the Form
  3. Support for device independence
  4. Standardized, powerful set of pre-defined declarative event handlers covering use cases. This introduces semantics and therefore assists accessibility.
  5. The introduction of standardized events that reduce the need for specialized JavaScript.

The first benefit is less JavaScript. XForms provides a common set of available actions that can be performed on a form. XForms is a delarative markup. The engineers who designed XForms took the standard use cases for how we create forms and built in support for the standard actions performed in a web form. From an accessibility perspective, this means these functions should be available from the browser independent of platform. So, rather than guess what an onclick does on a form, the user should be able to access the available functions and Notification events that are used in that specific form as defined by the XForms standard.

Interaction events:
- next and previous
- focus change
- help and hint
- refresh
- revalidate
- recalculate
- rebuild
- reset
- submit

You are also notified of events which actually occur during form operation. These types of events allow applications, like a screen reader, to provide alternative modalities such as speech.

Notification Events:

- Activate
- Value change
- Select and Deselect
- Scroll First, Scroll Last
- Insert, Delete
- Valid, Invalid
- FocusIn
- FocusOut
- ReadOnly, ReadWrite
- Required, Optional
- in-range, out-of-range
- submit-done
- submit-error


In addition to device independent events, XForms also provides another powerful feature by its model-view-controller design. XForms separates content from presentation while embedding relevant meta data in user interface constructs.

XForms provides the following meta data:
label for the user interface control
hint help text for the user interface control
help help text for the user interface control

XForms also encourages intent-based authoring by introducing these constructs:

group to group a set of user interface controls
switch , and case which can be used to hide or reveal logical groups of controls

case includes a Boolean attribute called selected that allows the group of controls within that case to become active. This is important to assistive technologies, such as a screen reader, which would monitor the document to determine when to speak selected elements. If none of the cases within the switch are selected, the first case in document order becomes active by default. How those elements are rendered visually is of no consequence. The semantics are preserved.

The use of a standard data model reduces extensive author enablement work in the area of adding accessibility properties to content. The model, when bound to XForms elements, may be used by the user agent to directly support the platform accessbility API as shown in this abridged table:


State
XForms  today User Agent Mapping (MSAA)
User Agent Mapping ATK (Gnome,KDE)
disabled
<select1>, <item>, <textarea>, <secret>, <input> STATE_SYSTEM_UNAVAILABLE


This is determined from relevant. If relevant is false this disabled shall be mapped
ATK:ATK_STATE_DISABLED
focusable
<select1>, <item>, <textarea>, <secret>, <input>, <choices>, <repeat>

STATE_SYSTEM_FOCUSABLE


This is determined from relevant. If relevant is false this disabled shall be mapped

ATK:ATK_STATE_FOCUSABLE
invalid
<select1>, <item>, <textarea>, <secret>, <input>

no mapping



ATK:ATK_STATE_INVALID

If the state is computed to be invalid, out-of-range, or required this value should be returned as true. In the case of MSAA the user agent should provide a specialized API to return its value. Alternatively, if the user agent provides a specialized API for XForms it may provide invalid(), outOfRange(),  or empty() (returns true when required but not available).  This information is computed from the instance data associated with the form element.
hidden
<select1>, <item>, <textarea>, <secret>, <input>, <choices>, <repeat>, <submit>, <trigger>

STATE_SYSTEM_INVISIBLE: true if hidden=true; Otherwise false

The user agent will need to set this to true when there is no binding to data in a form element. Elements with this state should not be made available to users.


ATK_STATE_VISIBLE : false if hidden=true.
Otherwise false

The user agent will need to set this to true when there is no binding to data in a form element. Elements with this state should not be made available to users.
multiline
<textarea>
no mapping. This is implied based on the textarea being representative of it role. ATK_STATE_MULTI_LINE

multiple
<select>, <repeat>
STATE_SYSTEM_EXTSELECTABLE

This attribute is implied based on the element type in XForms. It is up to the User Agent to do the implied mapping to the appropriate state.
ATK_STATE_MULTISELECTABLE

This attribute is implied based on the element type in XForms. It is up to the User Agent to do the implied mapping to the appropriate state.
optional
<select1>, <item>, <textarea>, <secret>, <input>, <choices>
no mapping

This attributed is determined from from the XForms model. The XForms implementation should provide a method either in the DOM or a specialized API since this value should be calculated as optional may be set to true or be evaluated based on an XPath. For example, a method of isOptional() is suggested.
no mapping

This attributed is determined from from the XForms model. The XForms implementation should provide a method either in the DOM or a specialized API since this value should be calculated as optional may be set to true or be evaluated based on an XPath. For example, a method of isOptional() is suggested.
readonly
<select1>, <item>, <textarea>, <secret>, <input>, <choices>

STATE_SYSTEM_READONLY

This attributed is determined from from the XForms model. The XForms implementation should provide a method either in the DOM or a specialized API since this value should be calculated as required may be set to true or be evaluated based on
an XPath. In the case of Windows this would be computed and added to the MSAA accessible states. In ATK
this would be computed and added to the AccessibleStateSet. On other systems a method could be provided
such as isReadOnly().
ATK_STATE_READONLY=inverse of readonly

This attributed is determined from from the XForms model. The XForms implementation should provide a method either in the DOM or a specialized API since this value should be calculated as required may be set to true or be evaluated based on
an XPath. In the case of Windows this would be computed and added to the MSAA accessible states. In ATK
this would be computed and added to the AccessibleStateSet. On other systems a method could be provided
such as isReadOnly().
required <select1>, <item>, <textarea>, <secret>, <input>, <choices> no mapping.

This attributed is determined from from the XForms model. The XForms implementation should provide a method either in the DOM or a specialized API since this value should be calculated as required may be set to true or be evaluated based on an XPath. For example, a method of isRequired() is suggested.
no mapping.

This attributed is determined from from the XForms model. The XForms implementation should provide a method either in the DOM or a specialized API since this value should be calculated as required may be set to true or be evaluated based on an XPath. For example, a method of isRequired() is suggested.
 secret
<secret STATE_SYSTEM_PROTECTED
ROLE_SYSTEM_TEXT

This is state is managed by the user agent. Use of secret as a tag implies secret.

ATK_ROLE_PASSWORD_TEXT

This is state is managed by the user agent. Use of secret as a tag implies secret.
selectable
<item>
STATE_SYSTEM_SELECTABLE

Note: <itemset> should be expanded by the user agent. An attribute of selectable should automatically be set by the user agent as long as it is bound.
ATK_STATE_SELECTABLE

Note: <itemset> should be expanded by the user agent. An attribute of selectable should automatically be set by the user agent as long as it is bound.
singleline <input> implied by <input> being representative of its role.

ATK_STATE_SINGLE_LINE

This saves authoring time by levering the XForms model to support accessibility directly. IBM is beginning development platform accessibility support for XForms in 2005 as a contribution to the open source community.

4.5.3 XHTML2

XHTML 2 is being desiging to build accessibility in from the ground floor. It is clear from the roadmap that some XHTML 2 accessibility features are borrowed through the use of name spaces to extend XHTML 1.1 through modularization. XHTML 2 incorporates the role attribute, document landmarks, XForms, XML Events, and a number of other accessibility features. It introduces an access element designed to address device independence, semantic navigation, and activation. It also benefits from the WAI GUI Role Taxonomy which is a cross-cutting  technology. The XHTML 2 specification is still in working draft form. A future list of the extensive accessibility features of XHTML 2 will be necessary.

4.6 User Agent Support

User agents are required to provide interoperable interfaces to assistive technology run times to produce a usable experience for users with a specific disability. A task of the roadmap will be to define requirements for UAAG. This is necessary provide a cohesion between user agent and author supplied state information.

User agents may need to adjust their user interface in response to state information stored in document elements. For example, an element having the focusable attribute set to true should result in a desktop user agent's placing the element in the keyboard tabbing order.

4.7 Assistive Technology Support

The WAI PF working group and IBM are working with assistive technologies to ensure support of the roadmap.

4.8 Authoring Tool Support

The WAI PF working group will be looking at ways to address authoring tools requirements for testing the accessibilitty of scripted web content. The development of authoring tools to assist the author in testing dynamic web content for accessibility is important. Today's accessibility fixers and checkers test to static web content. Accessibility properties will change during web page use. Web authoring tools should consider reusable accessible widgets and prompt for variable accessibility information such as title text. The use of RDF in the WAI GUI Role Taxonomy should be used to ensure which properties may be supported by which roles. W3C Authoring Tools Accessibility Guidelines 2.0 is developing requirements that will better address authors will creating accessible, dynamic, scripted web content.

4.9 Developer Guidelines and Implementations

At this time, IBM is doing much of the prototyping of this work. Sample markup has been contributed by IBM and may be found on the Mozilla accessibility web site. IBM is working on a best practices document and is contributing to WCAG 2.0 implementation techniques. UB Access is contributing much of the RDF design supporing the Role taxonomy work. At this time these guidelines are not yet made available.

4.10 Roadmap Deliverable Timeline

The following table is the technology deliverable Timeline for the Dynamic Accessible Web Content Roadmap.


Specfication WD 1.0
LC
CR Implementation Report PR REC
WAI States and Adaptable Properties Module 3Q05
4Q05
1Q06
2Q06
3Q06
TBD
WAI Role Taxonomy for Accessible Adaptable Applications 3Q05
4Q05
1Q06
2Q06
3Q06
TBD
XHTML2 Specificaton (includes XForms) 2Q05
TBD
TBD
TBD
TBD
TBD
Forward Looking Efforts:






XML Events with handler descriptions (not critical for XHTML) TBD
TBD
TBD
TBD
TBD
TBD
XForms 1.0





3Q03


5.0 Conclusions


This roadmap is designed to address the accessibility of dynamic, scripted, web content that may be rendered in today's browser while bridging to future declarative standards, such as XHTML2, in the W3C. The extensions being created for XHTML 1.X are intended to be cross-cutting. This roadmap clearly indicates that WCAG, ATAG, and UAAG should be in lock step for a common strategy. All groups are impacted and are interdependent. Additionally, close work with mainstream working groups, like HTML, CSS, and Device Independence will not only produce a working solution but it will produce a working solution that will have a broader impact than just XHTML.

6.0 Appendix

6.1 W3C Deliverables: New Specifications to Fill the Gaps

The following table indicates the specifications to track which address the gaps in the Roadmap.

Specfication XHTML 1.0, XHTML 1.1
XHTML 2.0
Accessible State Meta Data
WAI States and Adaptable Properties Module XHTML2 Specificaton (includes XForms)
TABINDEX Correction for focus
WAI States and Adaptable Properties Module All elements are focuseble in XHTML 2
Accessible Role Meta Data
WAI Role Taxonomy for Accessible Adaptable Applications WAI Role Taxonomy for Accessible Adaptable Applications
Role Attribute
Borrow from XHTML2: XHTML 2 Role Specification
XHTML2 Role Specification

6.2 References

CSS3 Basic User Interface Module

http://www.w3.org/TR/css3-ui/

CSS3 Selectors

http://www.w3.org/TR/css3-selectors/

Microsoft Active Accessibility (MSAA) state information

http://msdn.microsoft.com/library/default.asp?url=/library/en-us/msaa/msaaccrf_8y2b.asp

Gnome Accessibility Project (GAP) State and StateSets.

http://developer.gnome.org/projects/gap/tech-docs/at-spi-docs/at-spi-cspi-state-and-statesets.html

Apple Accessibility for Cocoa

http://developer.apple.com/documentation/Cocoa/Conceptual/Accessibility/index.html

Firefox DHTML Accessibility technology preview

http://www.mozilla.org/access/dhtml/

RDF Accessibility

http://ubaccess.com/rdf.html