System and method for enhancing resource accessibility

[0001] This application claims priority as a continuation-in-part of PCT/IL 03/00255, filed Mar. 27, 2003, which claims priority from U.S. provisional patent application Ser. No. 60/370,199, filed Apr. 5, 2002.

[0002] The present invention relates to rendering the content of a resource, and more particularly, is directed to creating a knowledge base relating to the content and formatting of the resource, and using the knowledge base to convert the original resource to a more accessible resource based on user needs such as device capability and human perception capability. The resource may be an Internet website, web content, media document or a multimedia document.

[0003] A problem with conventional websites is that neither the designer who designs their format and layout, nor the writer who decides on their content, provide for the vast range of devices and users that may access the websites. More specifically, devices have varying display sizes, varying audio capabilities and varying bandwidth connections. Users have varying abilities and disabilities, such as visual, auditory, perceptual, cognitive, learning, mobility, background knowledge, memory or other, which may depend on the user herself and/or characteristics of the user's environment, such as dim lighting, extremely bright lighting or ambient noise and so on.

[0004] Generally, prior art tools which address this situation are limited in the types of scenarios that can be accommodated, e.g., only lighting constraints or only screen element size.

[0005] In the United States, the most widely used accessibility guidelines are those specified for implementation of section 508 of the Rehabilitation Act of 1973, as amended, 29 U.S.C. 794d. Section 508 requires that when Federal agencies develop, procure, maintain, or use electronic and information technology, Federal employees with disabilities have access to and use of information and data that is comparable to the access and use by Federal employees who are not individuals with disabilities, unless an undue burden would be imposed on the agency. Section 508 also requires that individuals with disabilities, who are members of the public seeking information or services from a Federal agency, have access to and use of information and data that is comparable to that provided to the public who are not individuals with disabilities, unless an undue burden would be imposed on the agency. The guidelines are at http://www.access-board.gov/sec508guide/. The 508 guidelines are as follows:

• • [0006] (a) A text equivalent for every non-text element shall be provided (e.g., via “alt”, “longdesc”, or in element content).
  • [0007] (b) Equivalent alternatives for any multimedia presentation shall be synchronized with the presentation.
  • [0008] (c) Web pages shall be designed so that all information conveyed with color is also available without color, for example from context or markup.
  • [0009] (d) Documents shall be organized so they are readable without requiring an associated style sheet.
  • [0010] (e) Redundant text links shall be provided for each active region of a server-side image map.
  • [0011] (f) Client-side image maps shall be provided instead of server-side image maps except where the regions cannot be defined with an available geometric shape.
  • [0012] (g) Row and column headers shall be identified for data tables.
  • [0013] (h) Markup shall be used to associate data cells and header cells for data tables that have two or more logical levels of row or column headers.
  • [0014] (i) Frames shall be titled with text that facilitates frame identification and navigation.
  • [0015] (j) Pages shall be designed to avoid causing the screen to flicker with a frequency greater than 2 Hz and lower than 55 Hz.
  • [0016] (k) A text-only page, with equivalent information or functionality, shall be provided to make a web site comply with the provisions of this part, when compliance cannot be accomplished in any other way. The content of the text-only page shall be updated whenever the primary page changes.
  • [0017] (l) When pages utilize scripting languages to display content, or to create interface elements, the information provided by the script shall be identified with functional text that can be read by assistive technology.
  • [0018] (m) When a web page requires that an applet, plug-in or other application be present on the client system to interpret page content, the page must provide a link to a plug-in or applet that complies with §1194.21(a) through (l).
  • [0019] (n) When electronic forms are designed to be completed on-line, the form shall allow people using assistive technology to access the information, field elements, and functionality required for completion and submission of the form, including all directions and cues.
  • [0020] (o) A method shall be provided that permits users to skip repetitive navigation links.
  • [0021] (p) When a timed response is required, the user shall be alerted and given sufficient time to indicate more time is required.

[0022] The World-Wide Web Consortium (W3C) has published W3C Recommendation, dated May 5, 1999, for Web Content Accessibility Guidelines (WCAG) 1.0, available at http://www.w3.org/TR/WAI-WEBCONTENT/, explaining how to make web content accessible to people with disabilities. WCAG comprises guidelines, which are general goal statements, and checkpoints, which are specific things to do in support of the guidelines. Table 1 provides the current WCAG guidelines and checkpoints. Each checkpoint has a priority level assigned thereto.

TABLE 1
WCAG
No.	Topics	Priority	Checkpoint description
1	GUIDELINE		PROVIDE EQUIVALENT ALTERNATIVES TO
			AUDITORY AND VISUAL CONTENT.
			Provide content that, when presented to the user, conveys
			essentially the same function or purpose as auditory or
			visual content.
1.1	General	P1	Provide a text equivalent for every non-text element (e.g.,
			via “alt”, “longdesc”, or in element content). This
			includes: images, graphical representations of text
			(including symbols), image map regions, animations (e.g.,
			animated GIFs), applets and programmatic objects, ascii
			art, frames, scripts, images used as list bullets, spacers,
			graphical buttons, sounds (played with or without user
			interaction), stand-alone audio files, audio tracks of video,
			and video.
1.2	Images and	P1	Provide redundant text links for each active region of a
	image maps		server-side image map.
1.3	Images and	P1	Until user agents can automatically read aloud the text
	image maps		equivalent of a visual track, provide an auditory
			description of the important information of the visual track
			of a multimedia presentation.
1.4	Multimedia	P1	For any time-based multimedia presentation (e.g., a movie
			or animation), synchronize equivalent alternatives (e.g.,
			captions or auditory descriptions of the visual track) with
			the presentation.
1.5	Images and	P3	Until user agents render text equivalents for client-side
	image maps		image map links, provide redundant text links for each
			active region of a client-side image map.
2	GUIDELINE		DON'T RELY ON COLOR ALONE.
			Ensure that text and graphics are understandable when
			viewed without color.
2.1	General	P1	Ensure that all information conveyed with color is also
			available without color, for example from context or
			markup.
2.2	General	P2	Ensure that foreground and background color
			combinations provide sufficient contrast when viewed by
			someone having color deficits or when viewed on a black
			and white screen. [Priority 2 for images, Priority 3 for
			text].
3	GUIDELINE		USE MARKUP AND STYLE SHEETS AND DO SO
			PROPERLY.
			Mark up documents with the proper structural elements.
			Control presentation with style sheets rather than with
			presentation elements and attributes.
3.1	General	P2	Use markup languages when reasonable to convey
			information.
3.2	General	P2	Create documents that validate to published formal
			grammars.
3.3	General	P2	Use style sheets to control presentation.
3.4	General	P2	Use relative rather than absolute units in markup language
			attribute values and style sheet property values.
3.5	General	P2	Use mark up elements to convey document structure and
			use them according to specification.
3.6	General	P2	Mark up lists and list items properly.
3.7	General	P2	Mark up quotations. Do not use quotation markup for
			formatting effects such as indentation.
4	GUIDELINE		CLARIFY NATURAL LANGUAGE USAGE.
			Use markup that facilitates pronunciation or interpretation
			of abbreviated or foreign text.
4.1	General	P1	Clearly identify changes in the natural language of a
			document's text and any text equivalents (e.g., captions).
4.2	General	P3	Specify the expansion of each abbreviation or acronym in
			a document where it first occurs.
4.3	General	P3	Identify the primary natural language of a document.
5	GUIDELINE		CREATE TABLES THAT TRANSFORM
			GRACEFULLY.
			Ensure that tables have necessary markup to be
			transformed by accessible browsers and other user agents.
5.1	Tables	P1	For data tables, identify row and column headers.
5.2	Images and	P1	For data tables that have two or more logical levels of row
	image maps		or column headers, use markup to associate data cells and
			header cells.
5.3	Tables	P2	Do not use tables for layout unless the table makes sense
			when linearized. Otherwise, if the table does not make
			sense, provide an alternative equivalent (which may be a
			linearized version).
5.4	Tables	P2	If a table is used for layout, do not use any structural
			markup for the purpose of visual formatting.
5.5	Tables	P3	Provide summaries for tables.
5.6	Tables	P3	Provide abbreviations for header labels.
6	GUIDELINE		ENSURE THAT PAGES FEATURING NEW
			TECHNOLOGIES TRANSFORM GRACEFULLY.
			Ensure that pages are accessible even when newer
			technologies are not supported or are turned off.
6.1	General	P1	Organize documents so they may be read without style
			sheets. For example, when an HTML document is
			rendered without associated style sheets, it must still be
			possible to read the document.
6.2	General	P1	Ensure that equivalents for dynamic content are updated
			when the dynamic content changes.
6.3	Applets and	P1	Ensure that pages are usable when scripts, applets, or
	scripts		other programmatic objects are turned off or not
			supported. If this is not possible, provide equivalent
			information on an alternative accessible page.
6.4	Applets and	P2	For scripts and applets, ensure that event handlers are
	scripts		input device-independent.
6.5	General	P2	Ensure that dynamic content is accessible or provide an
			alternative presentation or page.
7	GUIDELINE		ENSURE USER CONTROL OF TIME-SENSITIVE
			CONTENT CHANGES.
			Ensure that moving, blinking, scrolling or auto-updating
			object or pages may be paused or stopped.
7.1	General	P1	Until user agents allow users to control flickering, avoid
			causing the screen to flicker.
7.2	General	P2	Until user agents allow users to control blinking, avoid
			causing content to blink (i.e., change presentation at a
			regular rate, such as turning on and off).
7.3	Applets and	P2	Until user agents allow users to freeze moving content,
	scripts		avoid movement in pages.
7.4	General	P2	Until user agents provide the ability to stop the refresh, do
			not create periodically auto-refreshing pages.
7.5	General	P2	Until user agents provide the ability to stop auto-redirect,
			do not use markup to redirect pages automatically.
			Instead, configure the server to perform redirects.
8	GUIDELINE		ENSURE DIRECT ACCESSIBILITY OF EMBEDDED
			USER INTERFACES.
			Ensure that the user interface follows principles of
			accessible design: device-independent access to
			functionality, keyboard operability, self-voicing, etc.
8.1	Applets and	P2	Make programmatic elements such as scripts and applets
	scripts		directly accessible or compatible with assistive
			technologies [Priority 1 if functionality is important and
			not presented elsewhere, otherwise Priority 2.]
9	GUIDELINE		DESIGN FOR DEVICE-INDEPENDENCE.
			Use features that enable activation of page elements via a
			variety of input devices.
9.1	Images and	P1	Provide client-side image maps instead of server-side
	image maps		image maps except where the regions cannot be defined
			with an available geometric shape.
9.2	Applets and	P2	Ensure that any element that has its own interface can be
	scripts		operated in a device-independent manner.
9.3	Applets and	P2	For scripts, specify logical event handlers rather than
	scripts		device-dependent event handlers.
9.4	General	P3	Create a logical tab order through links, form controls, and
			objects.
9.5	General	P3	Provide keyboard shortcuts to important links (including
			those in client-side image maps), form controls, and
			groups of form controls.
10	GUIDELINE		USE INTERIM SOLUTIONS.
			Use interim accessibility solutions so that assistive
			technologies and older browsers will operate correctly.
10.1	General	P2	not cause pop-ups or other windows to appear and do not
			change the current window without informing the user.
10.2	Forms	P2	Until user agents support explicit associations between
			labels and form controls, for all form controls with
			implicitly associated labels, ensure that the label is
			properly positioned.
10.3	Tables	P3	Until user agents (including assistive technologies) render
			side-by-side text correctly, provide a linear text alternative
			(on the current page or some other) for all tables that lay
			out text in parallel, word-wrapped columns.
10.4	Forms	P3	Until user agents handle empty controls correctly, include
			default, place-holding characters in edit boxes and text
			areas.
10.5	General	P3	Until user agents (including assistive technologies) render
			adjacent links distinctly, include non-link, printable
			characters (surrounded by spaces) between adjacent links.
11	GUIDELINE		USE W3C TECHNOLOGIES AND GUIDELINES.
			Use W3C technologies (according to specification) and
			follow accessibility guidelines. Where it is not possible to
			use a W3C technology, or doing so results in material that
			does not transform gracefully, provide an alternative
			version of the content that is accessible.
11.1	General	P2	Use W3C technologies when they are available and
			appropriate for a task and use the latest versions when
			supported.
11.2	General	P2	Avoid deprecated features of W3C technologies.
11.3	General	P3	Provide information so that users may receive documents
			according to their preferences (e.g., language, content
			type, etc.)
11.4	And if all	P1	OR provide a link to an alternative page that uses W3C
	else fails		technologies, is accessible, has equivalent information (or
			functionality), and is updated as often as the inaccessible
			(original) page.
12	GUIDELINE		PROVIDE CONTEXT AND ORIENTATION
			INFORMATION.
			Provide context and orientation information to help users
			understand complex pages or elements.
12.1	Frames	P1	Title each frame to facilitate frame identification and
			navigation.
12.2	Frames	P2	Describe the purpose of frames and how frames relate to
			each other if it is not obvious by frame titles alone.
12.3	General	P2	Divide large blocks of information into more manageable
			groups where natural and appropriate.
12.4	Forms	P2	Associate labels explicitly with their controls.
13	GUIDELINE		PROVIDE CLEAR NAVIGATION MECHANISMS.
			Provide clear and consistent navigation mechanisms -
			orientation information, navigation bars, a site map, etc. -
			to increase the likelihood that a person will find what they
			are looking for at a site.
13.1	General	P2	Clearly identify the target of each link.
13.2	General	P2	Provide metadata to add semantic information to pages
			and sites.
13.3	General	P2	Provide information about the general layout of a site
			(e.g., a site map or table of contents).
13.4	General	P2	Use navigation mechanisms in a consistent manner.
13.5	General	P3	Provide navigation bars to highlight and give access to the
			navigation mechanism.
13.6	General	P3	Group related links, identify the group (for user agents),
			and, until user agents do so, provide a way to bypass the
			group.
13.7	General	P3	With the client only - If search functions are provided,
			enable different types of searches for different skill levels
			and preferences.
13.8	General	P3	Place distinguishing information at the beginning of
			headings, paragraphs, lists, etc.
13.9	General	P3	Provide information about document collections (i.e.,
			documents comprising multiple pages.).
13.10	General	P3	Provide a means to skip over multi-line ASCII art.
14	GUIDELINE		ENSURE THAT DOCUMENTS ARE CLEAR AND
			SIMPLE.
			Ensure that documents are clear and simple so they may
			be more easily understood.
14.1	General	P1	Use the clearest and simplest language appropriate for a
			site's content.
14.2	General	P3	Supplement text with graphic or auditory presentations
			where they will facilitate comprehension of the page.
14.3	General	P3	Create a style of presentation that is consistent across
			pages.

[0023] Unfortunately, many websites do not follow these guidelines, either through ignorance, indifference or insufficient resources to ensure compliance.

[0024] Another problem is that the guidelines are intentionally limited, since the guideline creators focus on a subset of everything that could be done, to avoid frightening authors with a guideline that appears unduly onerous to comply with.

[0025] Accordingly, there is a need to improve the accessibility of resources without burdening the resource designers and writers; resources include websites, web content, multimedia files, complicated documents and books intended for electronic viewing.

[0026] In accordance with an aspect of this invention, there are provided a method of and a system for enhancing resource accessibility. Using a computer, a resource knowledge base is applied to the resource to generate a converted resource, the resource knowledge base providing information about the resource that is absent from or implicit in the resource, and the converted resource is provided to a user.

[0027] In accordance with another aspect of this invention, a resource knowledge base is created by using a computer to apply rules representing knowledge gap situations to a resource to identify knowledge gaps, and obtaining knowledge to fill the knowledge gaps, the obtained knowledge forming the resource knowledge base.

[0028] In accordance with a further aspect of this invention, a method of converting a resource from one format to at least one other target format comprises using a computer to apply at least one mapping knowledge base to the resource to generate at least one mapped resource, each mapping knowledge base providing information about the meaning of information whose format is to be converted so that the mapped resource is properly represented in the target format, and providing the at least one mapped resource to a user.

[0029] It is not intended that the invention be summarized here in its entirety. Rather, further features, aspects and advantages of the invention are set forth in or are apparent from the following description and drawings.

[0030] FIG. 1A is a diagram showing the configuration of a prior art rendering system;

[0031] FIG. 1B is a chart showing the main elements involved in a prior art page rendering;

[0032] FIG. 2A is a chart showing processing activity in a prior art accessibility checker program;

[0033] FIG. 2B is a chart showing processing activity in an accessibility compliance program;

[0034] FIGS. 3A-3B are diagrams showing configurations in which the present invention is applied;

[0035] FIG. 4 is a chart showing the components of an annotation wizard;

[0036] FIG. 5 is a chart showing how manually specified knowledge gaps are compiled into executable transformation programs;

[0037] FIG. 6 is a diagram showing a data structure for a knowledge gap file;

[0038] FIG. 7 is a chart showing the activity involved in creating a resource knowledge base;

[0039] FIGS. 8-9 are diagrams respectively showing data structures for intermediate and final knowledge bases;

[0040] FIGS. 10A-10C are charts showing the main elements involved in resource conversion according to the embodiments of the present invention;

[0041] FIG. 11 is a flowchart showing set-up of a user preferences file;

[0042] FIG. 12 is a diagram showing a data structure for a user preferences file;

[0043] FIGS. 13A-13F are a flowchart showing how a resource knowledge base is created for a resource;

[0044] FIGS. 14A-14C are a flowchart showing how a resource is converted to an accessible resource;

[0045] FIG. 15A is a chart showing the screen layout of an editor interface; and

[0046] FIG. 15B is a chart showing an image and dialog presented to the editor.

[0047] Software for converting information from one format to another is well known. In the area of websites, a rendering engine converts a web page from its conventional form to a reformatted form based on user preferences.

[0048] FIG. 1A depicts the environment of a conventional rendering engine. A user has personal computer 10 or other access device, such as a personal digital assistant (PDA), coupled to Internet 20. Using typical hypertext transfer protocol (http), the user sends a page request via Internet 20 to server 30, and server 30 responds with a page to be presented on the user's access device. Third party server 50 is also coupled to Internet 20.

[0049] Page rendering engine 35, comprising software executing on server 30, receives the page request, retrieves the original hypertext markup language (HTML) web page, and converts its format according to user preferences, which are stored at one or more of the user's access device, shown as user preferences 15, the server's data storage facility, shown as user preferences 44, and/or a third party website, shown as user preferences 55. The page rendering engine sends the reformatted web page to the user in response to the user's page request.

[0050] Conventional user preferences include user specified information, such as font size, device specific information, such as number of pixels in the display.

[0051] FIG. 1B is a diagram showing the above-described activity. Page 42 of a website is an input to page rendering engine 35. User preferences 15 are also an input to page rendering engine 35, which reformats page 42 in accordance with user preferences 15 to produce reformatted page 60.

[0052] A problem with the conventional rendering system is that it is completely unresponsive to the meaning of the website content. For example, if a bulleted list of items has red bullets indicating sale items, a conventionally rendered page for a user with a black-and-white display will only indicate that the bullet is red; the fact that the item is on sale will be transparent to the rendering. However, the fact that the item is on sale is probably much more important to the user than the color of the bullet. Another problem with the conventional rendering system is that to accommodate each new user scenario (learning disabled, flashing images aversion, forms of dyslexia), the rendering system itself and/or the resource creator's content must be revised.

[0053] FIG. 2A is a chart showing processing activity in a prior art accessibility checker program, specifically, the Bobby program developed by Center for Applied Special Technology (CAST), supported by Watchfire in Waltham, Mass., and available at http://bobby.watchfire.com/bobby/html/en/indexjsp.

[0054] Bobby is a comprehensive web accessibility software tool designed to help expose and repair barriers to accessibility and encourage compliance with existing accessibility guidelines. Bobby tests for compliance with government standards, including the U.S. Government's Section 508. It offers prioritized suggestions based on the Web Content Accessibility Guidelines provided by the World Wide Web Consortium's (W3C) Web Access Initiative. Bobby allows developers to test web pages and generate summary reports highlighting critical accessibility issues before posting content to live servers. Bobby tests web pages using the guidelines established by the World Wide Web Consortium's (W3C) Web Access Initiative (WAI), as well as Section 508 guidelines from the Architectural and Transportation Barriers Compliance Board (Access Board) of the U.S. Federal Government.

[0055] Bobby is run by a user on a web directory that includes hypertext markup language (HTML) files. At step 65, the Bobby program searches a resource for accessibility violations using a set of patterns; if the code for a resource matches the patterns, then Bobby determines that an accessibility standards violation exists. At step 70, Bobby advises its user, an editor, that a violation has been found. At step 75, the user manually changes the resource to comply with the accessibility standards. This process repeats until there are no accessibility violations detected in the resource. Bobby is not completely accurate since it does not properly evaluate the content of various elements, such as alt-tags.

[0056] FIG. 2B is a chart showing processing activity in accessibility compliance programs, specifically, ACCVERIFY and ACCMONITOR from HiSoftware Solutions, available at http://www.hisoftware.com/access/vlndex.html. These accessibility compliance programs operate similarly to Bobby, but include a graphic interface that accepts fixes for accessibility violations and automatically revises the HTML code for the original resource to create a revised, accessible resource.

[0057] At step 80, the accessibility compliane program searches a resource for accessibility violations using a set of patterns; if the code for a resource matches the patterns, then the program determines that an accessibility standards violation exists. At step 85, the program advises its user, an editor, that a violation has been found. At step 90, the program receives a change from the editor that may enable the resource to comply with the accessibility standards. That is, although the editor supplies a change, nothing checks whether the editor's change actually cures the accessibility violation. Since few editors are experts on accessibility issues, it is likely that most editors supply inadequate or inappropriate information for at least some of their edits. At step 95, the program revises the resource to incorporate the editor's change. This process repeats until there are no accessibility violations detected in the resource.

[0058] Drawbacks of the above-described programs include:

• • [0059] The programs are driven by a list of accessibility violation patterns. If a problem is not one of the predefined patterns, it escapes the program's attention;
  • [0060] The programs merely describe instances where the resource includes an instance of a violation pattern, that is, no guidance is given to the editor as to how to best fix the violation. If the editor does not understand accessibility issues, then the editor's change is frequently inadequate or inappropriate;
  • [0061] The programs require that the original resource be changed to conform to accessibility guidelines. However, resource owners are typically very nervous about changes to the original resource, particularly since the editor doing the accessibility compliance may not understand why things were designed in the original resource;
  • [0062] The programs result in only one accessibility-compliant version of the resource. However, users that are differently disabled may require different versions: a deaf person requires all visual, a blind person prefers more text over images, a learning-disabled person usually prefers more images than text and so on. Thus, the end result of the programs will be suboptimal for almost all disabled users;
  • [0063] There is an inherent conflict between optimizing a resource for accessibility by various parties and providing a highly stylized design for a website; thus, the programs may interfere with the look and feel of the website intended by the author;
  • [0064] The programs have no way of evaluating the overall meaning or importance of a resource; and
  • [0065] The programs have limited ability to restructure a resource to improve its accessibility, either by simplifying the existing structure or by adding additional structure such as an index. The programs' restructuring typically depends on the resource being configured for restructuring, such as using XHTML headers rather than embedded font description information.

[0066] Thus, there is enormous room to improve the functionality of programs that improve the accessibility of resources.

[0067] A resource accessibility engine according to the present invention uses a resource-specific knowledge base and user preferences to convert a resource into an improved accessibility resource. Examples of a resource include a website, document, webpage image file, multimedia file, auditory file or any other text and/or non-text record. The resource-specific knowledge includes content and formatting information which reduces ambiguities, translates implied information into explicit information and improves the accessibility of the resource content for persons having perceptual disabilities/preferences and/or presentation devices with limited capabilities and other devices such as knowledge processing systems, translation engines and middleware. The user preferences represent the physical capabilities of the user's access device, and the user's semantic and personal preferences for how content should be displayed.

[0068] An editor uses an annotation wizard to create the resource-specific knowledge base based on the original resource. The annotation wizard uses rules for guiding the editor through providing knowledge, specifically, providing explicit knowledge that is missing, converting implicitly provided knowledge into explicit knowledge, and confirming that the annotation wizard's rules are properly interpreting the meaning of content in the resource.

[0069] Advantages of programs according to the present invention include:

• • [0070] The present programs are driven by levels of information organization to engage an editor in a dialog about what is meant. Accordingly, everything not sufficiently understood is presented to the editor for review. This approach is fundamentally akin to “make sure everything is good enough”, and is much more powerful than the approach of the above-described programs of “find things known to be bad”.
  • [0071] The present programs guide the editor in at least the following ways:
    • [0072] the present programs often guess what the correct information is, so that the editor can simply agree, or has a more directed way of providing the correct information; and
    • [0073] the present programs prompt the editor for meaning at a variety of levels, for example, a description of an image might be “c inside a circle” while the meaning of the image is “there exists artistic expression protected by copyright”; and
    • [0074] the present programs prompt the editor for implicit meaning, such as is conveyed by formatting information;
  • [0075] The present programs do not change the original resource. Instead, the present programs enable the user to dynamically create a version of the original resource best suited to the user's perceptual abilities, while leaving the original resource unchanged, which minimizes anxiety on the part of the resource owner;
  • [0076] The present programs produce a converted resource in accordance with the user's preferences, referred to herein as alternative renderings. Thus, the end result of the present programs will be optimal for each user's perceptual abilities;
  • [0077] The present programs have multiple ways of evaluating the overall meaning or importance of a resource:
    • [0078] directly asking the editor, usually via a series of simple questions;
    • [0079] guessing, and asking the editor to evaluate the correctness of the guess;
    • [0080] guessing with a sufficiently high degree of certainty that editor confirmation is not needed, although the editor is always welcome to edit;
    • [0081] giving the editor multiple opportunities to describe a single portion of the resource, spurring the editor to think more carefully about the portion and thus provide more information;
    • [0082] the present programs assign roles, defined in the RDF statements, to portions of the resource, which helps organize the editor's description of the portions. A role represents how an element or section of a resource is used within the context of the resource and the function it performs. A role is expressed in computer-comprehensible form so they are inherently extendable, and new roles can be derived from existing roles, and the relationships between roles can also be defined in a programmatically understandable way. This is useful in groups that need their own type of content, e.g., an educational community may have a “tutor” button. The role of tutor can be derived from the role of help, as tutoring is a special case of helping. Rendering agents that do not know how to deal with the tutor role can deal with it as a help role. Specialized rendering agents will be able to properly use the special meaning of tutor as compared to help;
  • [0083] The present programs can change the structure of the converted resource relative to the original resource by providing additional structure such as a intra-page map showing how the portions of a long page are related, or an inter-page map showing how the pages of a resource are related;
  • [0084] The present programs can substantially change content in alternative renderings, in a way that would be unacceptable to the resource owner if the changes were applied to the original resource; these substantial changes require understanding the meaning of the resource, the relative importance of resource elements and similar concepts. Examples of this type of rendering include replacing text with symbols and simple words, and removing or hiding less important content, replacing ambiguous wording such as predicates and confusing word ambiguities with clear terms, providing extra help or text when it aids comprehension;
  • [0085] The present programs enable new types of transformations to be employed to create new types of renderings without requiring any new knowledge from the editor or user. For example, if a new method of displaying complex content to blind persons becomes available, the new method can be applied to existing resources based on the resource's knowledge base;
  • [0086] The present programs enable a third party to make a resource accessible even if the third party is unable to modify the original resource due to the non-invasive nature of the present programs;
  • [0087] The present programs re-use knowledge supplied by an editor to minimize the burden on the editor. For example, an editor needs to enter the expansion of an acronym only once, and thereafter the program uses the already-supplied definition instead of asking the editor to define the acronym again.

[0088] The following terms are used herein: a “user” is someone who wants a resource presented to them, preferably in accessible form in view of the user's perceptual preferences and abilities. An “author” is someone who created a resource. An “editor” is someone who interacts with an annotation wizard according to the present invention to provide information so that the resource can be presented in accessible form; the editor may, but need not be, the same person as the author.

[0089] The present invention uses techniques from the so-called semantic web to add knowledge to resources so that programs can better interact with the resources:

• • [0090] unique fragment identifiers or resource identifiers such as uniform resource identifiers (URIs) identify resources in general, both resources that are network retrievable and have a uniform resource locator (URL), and resources that are not network retrievable such as a person or a tangible thing. Unlike traditional RDF, discussed below, the invention is not limited in its use of identifiers to extensible markup language (XML) compatible identifiers; for example, by mapping cascaded style sheets to XML, the mapped cascaded style sheets can be identified. In other words, a mechanism for pointing is needed, and the mechanism can be other than XML;
  • [0091] resource description framework (RDF), discussed below, is used for capturing meaning and relationship in content;
  • [0092] an ontology, or list of terms used in RDF statements that specify a variety of relationship among data elements and ways of making logical inferences among them; and
  • [0093] An optional XML based platform for converted resource that contains tags and attributes for adding extra knowledge from the RDF statements to the XHTML or similar resource.

[0094] FIGS. 3A-3B are diagrams showing example configurations in which the present invention is applied. FIG. 3A shows a configuration involving the Internet, while FIG. 3B shows a standalone configuration.

[0095] FIG. 3A shows Internet 100 coupled to user personal computer (pc) 110, third party servers 120, 130 and accessibility server 150. A personal computer is a general purpose computer having an operating system and a web browser.

[0096] Server 150 is a general purpose computer or set of computers having storage 160 coupled thereto. Storage 160 is one or more high volume storage devices employing magnetic, optical, or any other suitable devices. Editor pc 170 is also coupled to storage 160. Any suitable coupling technique may be used, for example, a local area network may connect server 150, storage 160 and editor pc 170.

[0097] Server 150 functions to execute software programs, in particular, accessibility engine 152, annotation wizard 154 and user preferences setup program 156, as well as conventional web-server programs (not shown). Annotation wizard 154 is used by an editor to set up a knowledge base for a resource, so that accessibility engine 152 can apply the knowledge base to the resource to create a converted resource having improved accessibility. Annotation wizard 154 is used during a set-up phase. Accessibility engine 152 is called by annotation wizard 154 during the set-up phase, discussed below. Accessibility engine 152 is used by a user either dynamically, in real time, or at a preparation time, depending on the nature of the resource and user convenience. The exact locations of wizard 154 and engine 152 are not important. For example, the accessibility engine could be located on the user's pc, as is the case with accessibility engine 185 in FIG. 3B; and the annotation wizard can be located at editor pc 170, as is the case with annotation wizard 175, or on third party server 120, as is the case with annotation wizard 125.

[0098] User preferences setup program 156 is needed only if a user preferences file must be setup. In other cases, user preferences setup program 156 is omitted.

[0099] Storage 160 functions to store a resource—such as website 162—comprising pages 163 of hypertext markup language (HTML) code, resource knowledge base 164 relating to a specific resource such as website 162, user preferences 166, and scenario-specific style sheet rules 168. The exact locations of the resource, knowledge base, user preferences and scenario-specific style sheet rules are not important. For example, the resource could be at a third party server or distributed to users on a recording medium, as discussed with regard to FIG. 3B; the resource knowledge base could be on a third party server, as is the case with resource kb 144; and the user preferences can be on a third party server, as is the case with user preferences 142, or on user pc 100, as is the case with user preferences 175. More specifically, the user preferences can be an explicit file, included in a user's profile that is part of another party's data, or dynamically expressed through selections such as from a drop-down menu. The scenario-specific style sheet rules are typically proximate to an accessibility engine. FIG. 3A shows user pc 110 having scenario-specific style sheet rules 117; in other embodiments, user pc 110 does not have scenario-specific style sheet rules at its location. Optionally, a resource could have its own scenario-specific style sheet rules.

[0100] FIG. 3B corresponds to a situation where a content provider distributes recording medium 190 such as a compact disc (CD) having recorded thereon resource 192, such as a multimedia book, resource knowledge base 192 and an executable version of accessibility engine 185 that is loaded into user pc 180. As used herein, a resource is “local” to a user when the resource is accessible directly from the user's computer, without using the Internet. Thus, if a user uses a resource on a local area network or virtual private network that is private to an enterprise, the resource is local to the user.

[0101] FIGS. 4-9 describe creation of the resource knowledge base, and are discussed below.

[0102] As shown in FIG. 10A, accessibility engine 152 (or accessibility engine 185 in the case of FIG. 3B) uses as inputs website 162 (or resource 190 in the case of FIG. 3B), resource knowledge base 164 (or resource knowledge base 192 in the case of FIG. 3B) and user preferences file 166 (chosen dynamically, rather than from a file, in the case of FIG. 3B), to create converted resource 192 that includes knowledge necessary for proper presentation to the user in accordance with the user's perceptual preferences and/or end device presentation capabilities. Converted resource 192 is then presented to the user by the user's presentation device (user pc 180, in the case of FIG. 3B).

[0103] In some embodiments, instead of the knowledge base being directed to a specific resource, the knowledge base can be directed to how to map from one format to another; in this case, the knowledge base is referred to as a “mapping kb”. Examples of format pairs include WORD to WORDPERFECT, HTML to XML, Flex or Flash (www.macromedia.com) to Xforms (www.w3c.org) and so on.

[0104] FIG. 10B shows resource 162 and mapping kb 165 supplied to accessibility engine 152, which functions to create mapped resource 193 by applying the mapping information in mapping kb 165 to the elements of resource 162.

[0105] The resource knowledge and the mapping knowledge can be combined sequentially. FIG. 10C shows resource 162 and resource kb 164 supplied to accessibility engine 152 which functions to create converted resource 192. Next, a variety of mappings represented by mapping kb 165A . . . mapping kb 165N are supplied with respective copies of converted resource 192 to instances of accessibility engine 152 which functions to create respective mapped resources 193A . . . 193N.

[0106] An example of using this flexibility is as follows. A newsfeed really simple syndication (RSS) provider can provide a knowledge base accompanying the RSS feed, so that any content provider incorporating the RSS feed can use the feed in alternative renderings or in an alternative form.

[0107] The process of creating a knowledge base will now be discussed. Although an automated process is described, it will be appreciated that a knowledge base can be created manually or through a different process such as integration with resource authoring tools.

[0108] FIG. 4 is a chart showing the components of an annotation wizard: transformations programs 210, editor interface 220 and extracter 230.

[0109] Transformations programs 210 are fact-specific programs incorporating expertise about how to analyze different resources, identify knowledge gaps that indicate missing information and make implicit information into explicit information. Transformations programs 210 assume that human editors are unaware of various accessibility guidelines and the problems encountered by people lacking “normal” perceptual abilities and/or having disabilities, so transformation programs 210 function to meticulously comb through resources to find situations, referred to herein as knowledge gaps, where meaning can be readily augmented by a dialog with a human editor, and to collect the knowledge into a resource knowledge base. Using a knowledge base to address the knowledge gaps is an important aspect of how the present invention is able to convert a resource to convey clearer meaning in the resource to a user.

[0110] Editor interface 220 functions to present dialog regarding the output of transformations programs 210 to a human editor. Transformations programs 210 may operate in parallel, both for their own program modules and on different portions of a resource, so processing time is reduced. Additionally, the dialogs often present both resource code and screen displays corresponding to the resource code to an editor; editor interface 220 is responsible for managing the information flow to and from the human editor.

[0111] Editor interface includes lexical analyzer 225, interactivity analyzer 227 and complexity analyzer 229. Each analyzer can be used in standalone form. In some embodiments, each analyzer is a separate “plug-in” for another program such as a third-party website editing tool.

[0112] Lexical analyzer 225 uses cascaded lexicons to define the meaning of a word. A lexicon is a file having at least one word and an associated meaning for the word. Generally, there is a default lexicon, and optional editor-indicated special lexicons. Any number of special lexicons can be cascaded. The editor specifies the priority of the cascaded lexicons, which may differ by location in the resource. Editor interface 220 enables the editor to place a cursor over words in the resource and see the meaning assigned to the word; if the editor wishes to override a meaning, or supply a missing meaning, the editor's knowledge is captured as a new lexicon entry.

[0113] Lexicons can be selected by the editor based on the intended user, such as age group, educational level, and so on.

[0114] Symbol lexicons function in similar manner. An example of a symbol is an “i” inside a circle, meaning “information (help) is available here”. Another example of a symbol is a “c” inside a circle, meaning “protected by copyright”. Symbol lexicons are particularly useful for autistic children, who may have a vocabulary of 200 words and 500 symbols.

[0115] Interactivity analyzer 227 helps convert interactive content, such as Macromedia Flash files, to accessible content. Editor interface 220 collects at least the following information from the resource and the editor:

• • [0116] Element integrity (what is a page, what is an element);
  • [0117] Conditions (time, state—like selected, values);
  • [0118] Data types/role of data and form controls (like Date Email . . . );
  • [0119] Action types/role of data and form controls (Submit Validate . . . );
  • [0120] Hierarchal information (like Heading Label . . . );
  • [0121] Other relationships and roles (help, glossary);
  • [0122] Event information ( Focus, Mouse click . . . );
  • [0123] Text equivalencies;
  • [0124] Presentational information;
    and the interaction between them.

[0125] Complexity analyzer 229 serves to simplify complicated text. Complexity analyzer 229 parses the resource using a simple language dictionary; if there are difficult words, analyzer 229 suggests simpler words to the editor, and adds the editor's response to resource kb 164. This allows an editor to see where simple words have been used, and when they have not. When a simple word has not been used the editor can supply a simple alternative or glossary entry. Complexity analyzer 229 parses the resource for long sentences and complex word and sentence structures and the like, suggests areas for simplification to the editor, and adds the editor's response to resource kb 164. Complexity analyzer 229 parses the resource for long paragraphs that require high comprehension ability, and if they exist, suggests areas for simplification to the editor, and adds the editor's response to resource kb 164.

[0126] FIG. 15A is a chart showing the screen layout of an editor interface. Display 950 includes window 952 for presenting an element for which knowledge is being obtained from the editor, window 954 including hyperlinks for providing more detail about the element in window 952, window 956 for providing text from annotation wizard 154 to the editor, and window 958 for receiving input from the editor.

[0127] FIG. 15B is a chart showing an instance of image and dialog presented to the editor. Display 957 includes window 953 showing an image for which knowledge is being obtained. Window 957 shows dialog from annotation wizard 154 to the editor. Window 959 provides sequentially presented “radio buttons” for the editor to answer yes or no to the dialog questions, and provides a text box for free-form text entry. In this example, the editor has entered text describing the image as “Word “Cobras” with cartoon drawing of a cobra, the “o” is a drawing of a soccer ball.” In a subsequent dialog screen, the editor provides the additional information that this image is the logo for a soccer team having the team name is “Cobras”. The sequence of dialog depends on the editor's input. In this way, the questions guide the editor into providing the correct knowledge.

[0128] Extracter 230 functions to convert the intermediate form of a knowledge base, used by annotation wizard 154, into a final form that is more efficiently used by an accessibility engine.

[0129] FIG. 5 is a chart showing how manually specified knowledge gaps are compiled into transformation programs 210.

[0130] A programmer manually creates and edits knowledge gap file 200 using conventional text editing tools. Knowledge gap file 200 is organized into multiple levels, such as eight levels; in different embodiments different organizations and numbers of levels are used. The levels of the knowledge gap file 200 are used for sequencing the discovery of knowledge.

[0131] Accessibility standards are typically organized into levels or priorities; knowledge gap file 200 accommodates many standards, so its organization does not necessarily directly correspond to the organization of any of the accessibility standards. When using annotation wizard 154, an editor can indicate that only knowledge needed to satisfy a certain accessibility compliance level or task should be obtained, to avoid the lengthy process of providing knowledge for all possible gaps.

[0132] The conformance levels of accessibility standards are distinct from and essentially independent of the levels of the knowledge gap file 200.

[0133] FIG. 6 is a diagram showing a data structure for knowledge gap file 200, which comprises knowledge gap records 222. Knowledge gap file 200 is also referred to as master transformations file (MTF) 220. Each knowledge gap record has the information shown in Table 2.

TABLE 2
field	meaning
number	A unique number also referred to as a transformation number, assigned to
	each knowledge gap record, also referred to as a transformation record.
tag	Indicates the HTML tag that this knowledge gap record relates to. For
	example, a knowledge gap record relating to an HTML image (“img”)
	tag would be applicable to the following HTML code:
	<img SRC=“images/bb1.gif” >
logic	Essentially, a comment for programmers reading the file, describing what
	the transformation record is trying to do, that is, what kind of conversion
	should be applied to the resource.
used for	Knowledge gap and accessibility issues that the knowledge to be
	obtained from the editor is used for.
level	The knowledge gap (transformations) records are organized into levels
	that determine processing order. This field indicates the level that the
	instant record corresponds to.
conformance	Indicates the various standards checkpoints and so on that correspond to
	the transformation in this knowledge gap record.
testing	If a report of knowledge gaps is to be produced, then the comment found
	in the testing field is what is included in the knowledge gap report if this
	type of knowledge gap is found.
links	Hyperlinks to information about the accessibility violation that this
	knowledge gap is relevant to.
separate	Separate transformation is a rule that indicates characteristics of a new
transformation(s)	occurrence of a situation; if this occurrence exists in the resource, then a
	knowledge gap exists each time that situation addressed by the
	knowledge gap record is found. In contrast, if no separate transformation
	is specified, then the knowledge supplied at the first occurrence of the
	situation is assumed to be useful in all other occurrences of the situation
	in the resource. That is, separate transformations indicate new
	knowledge gaps.
case	Indicates the specific situations that will trigger this transformation, that
	is, for a knowledge rule in the form IF (case) THEN (actions), the “case”
	part of the rule.
action	Indicates the specific actions for obtaining of knowledge that should be
	taken when this transformation is applied, that is, for a knowledge rule in
	the form IF (case) THEN (actions), the “actions” part of the rule.
	Actions typically are used to write RDF statements possibly with best
	guesses to encapsulate and encode missing knowledge.

[0134] An instance of knowledge gap record 222 is shown in Table 3. This particular record is concerned with a situation where there is a hyperlink indicated by an image, but no text or “alt” information is associated with the image. WCAG checkpoint 1.1 (see Table 1), inter alia, requires that where there is an image with no text or “alt” information, then a text equivalent must be provided. This record specifies that in this situation, a knowledge entry is created associating alternative text of value “go to /page title/” where /page title/ is the title of the linked-to page.

TABLE 3
<transformation>
<number>1.1.6</number>
<tag>img</tag>
<logic>locate images used as the content of a link. If no alt attribute and no link text are
provided for text equivalency, then provide useful Alternate content that directs the user to the
linked to page</logic>
<usedFor>A</usedFor>
<level>Transformations 2</level>
<conformance>
<WCAG01-p1/>
<WCAG01-p2/>
<WCAG01-guideline1/>
<WCAG01-checkpoint1.1/>
<Section508/>
<Section508-paragraph22a/>
<dependency-level2/>
</conformance>
<testing>
<en>Missing text equivalent. (Suggestion: Put the name of the page that you are linking to in
the “alt” tags.)</en>
</testing>
<links>
<WCAG01-text>http://www.w3.org/TR/WAI-WEBCONTENT-TECHS/#gl-provide-
equivalents</WCAG01-text>
<Section508-text>http://www.access-board.gov/sec508/guide/1194.22.htm#(a)</Section508-
text>
<WCAG01-examples>http://www.w3.org/TR/WCAG10-HTML-TECHS/#link-text-
images</WCAG01-examples>
</links>
<separateTransformation>
<description>an href of the ancestor A is unique</description>
<information action=“distinct” name=“href” type=“attribute” path=“ancestor::a/”/>
</separateTransformation>
<separateTransformation>
<description>a src of the image is unique</description>
<information type=“attribute” name=“src” action=“distinct”/>
</separateTransformation>
<case>
<description>an image is used as the content of a link and
neither link text nor alt attribute is provided
</description>
<if type=“tag” name=“img” path=“//a/”/>
<if action=“missing” name=“alt” path=“” type=“attribute”/>
<if action=“empty” type=“content” path=“ancestor::a”/>
</case>
<action>
<description>alternate text with value “go to URI” of the page</description>
<then action=“Altext” path=“” type=“attribute”>
<literal> go to </literal>
<pattern>URL#pointer (descendant::title)</pattern>
</then>
</action>
</transformation>

[0135] Appendix A describes the knowledge gap records used in an embodiment of the invention. Table 4 shows the information provided for each of the knowledge gap records in Appendix A.

TABLE 4
Field	Description
Transformation	A unique number assigned to each knowledge gap record, also referred to as
Number	a transformation record.
Dependency	The knowledge gap (transformations) records are organized into levels. This
	field indicates the level that the instant record corresponds to.
Knowledge gap	A description of the knowledge that is missing from the original resource so
	that the contents of the resource can be converted in a way that retains the
	meaning intended in the original.
Logic	An example of how the logic, described in the about field, might refer to an
	HTML document.
Used for	Knowledge gap and accessibility issues that the knowledge to be obtained
	from the editor is used for.
About	Description of what the transformation record is trying to do, that is, what
	kind of conversion should be applied to the resource.
Best guess	To make knowledge capture faster, the annotation wizard offers best guesses
	to the editor. The editor can change the best guesses.
Editor supplied	“None” if there is high confidence that the best guess is correct. Otherwise, a
knowledge	description of the knowledge to be gleaned from the editor. In either case,
	the editor can always edit knowledge responsive to a knowledge gap.
Example case	Examples of occurrences that would trigger the instant transformation in an
(HTML	HTML document.
description)
Example use,	Example of how knowledge is applied to an HTML page to make a WCAG
WCAG	conformant page.
description

[0136] Knowledge gap compiler 205 converts knowledge gap file 200 into transformations programs 210. Knowledge gap compiler 205 operates on the entire set of knowledge gap records 222 to generate transformations programs 210.

[0137] Transformations programs 210 function to enable a computer to recognize knowledge gap situations in a file. The knowledge gap situations include situations where knowledge is simply missing and explicitly needs to be provided, situations where it appears there is implicit knowledge in the resource, and a human editor needs to be prompted to provide the knowledge, and situations where there is missing knowledge but enough context for the computer to guess what the knowledge is. The situations where the computer guesses knowledge are either “high confidence” situations where the editor need not confirm the guessed knowledge—but always may overwrite the guessed knowledge—or “other” situations where a human editor needs to confirm or correct the guess.

[0138] Transformations programs 210 are organized into levels corresponding to the levels of knowledge gap file 200. Each level is sometimes referred to herein as a separate transformations program for a respective level. For example, if there are eight levels, there is a set of transformations programs 211, 212, 213, . . . 218. However, each transformations program is actually a portion of an organic entity, the set of transformations programs 210.

[0139] Transformations programs 210 additionally function to create a “starter” resource knowledge base comprising records corresponding to each missing knowledge gap and best guesses as to knowledge to fill the knowledge gaps. The starter resource knowledge base is augmented with knowledge collected from the human editor. Each level of transformations program may use knowledge collected from a previous level in determining whether there is a knowledge gap. Accordingly, although some transformation programs may be executed in parallel, there are some transformations programs that must be executed according to a sequential schedule. There is a project plan file (not shown) used by annotation wizard 154 to ensure proper scheduling during parallel processing. The project plan file is created by a human programmer and reused for different resources of the same type; when a new level of processing is added, then the project plan is modified.

[0140] FIG. 7 is a chart showing the activity involved in creating a resource knowledge base. Annotation wizard 154 is invoked by a human editor to process resource 162, which serves as an initial resource. Transformations program level one 211 is executed on the initial resource to create resource knowledge base (kb) level one 231. After transformations program level one 211 has been run, annotation wizard 154 causes editor interface 220 to engage in a dialog with a human editor to obtain information and confirm guessed information.

[0141] Accessibility engine 152 is invoked by annotation wizard 154 to process the initial resource and resource kb level one 231 to generate resource level one 241, corresponding to the initial resource augmented with the knowledge provided by transformations program level one 211. Transformations program level two 212 is executed on resource level one 241 to create resource kb level two 232. After transformations program level two 212 has been run, annotation wizard 154 causes editor interface 220 to engage in a dialog with a human editor to obtain information and confirm guessed information.

[0142] Accessibility engine 152 is invoked by annotation wizard 154 to process the resource level one 241 and resource kb level two 232 to generate resource level two 242, corresponding to the initial resource augmented with the knowledge provided by transformations programs levels one and two 211, 212. Transformations program level three 213 is executed on resource level two 242 to create resource kb level three 233. After transformations program level three 213 has been run, annotation wizard 154 causes editor interface 220 to engage in a dialog with a human editor to obtain information and confirm guessed information.

[0143] Accessibility engine 152 is invoked by annotation wizard 154 to process the resource level two 242 and resource kb level three 233 to generate resource level three 243, corresponding to the initial resource augmented with the knowledge provided by transformations programs levels one, two and three 211, 212, 213. Transformations program level four 214 is executed on resource level three 243 to create resource kb level four 234. After transformations program level four 214 has been run, annotation wizard 154 causes editor interface 220 to engage in a dialog with a human editor to obtain information and confirm guessed information.

[0144] This processing sequence continues until the nth level of transformations program, in this example, transformations program level eight 218, is executed on resource level seven 247 to create resource kb level eight 238. Annotation wizard 154 causes editor interface 220 to engage in a dialog with a human editor to obtain information and confirm guessed information.

[0145] Extracter 230 then extracts relevant information from resource kb level eight 238, specifically, applicable graphs 262, to create a final resource knowledge base, in this example, resource kb 164. If a resource is configured with components, such as pages, groups of pages with a lot of common content, separate knowledge bases can be created for the respective components; this provides the advantage of enabling only a selected portion of a resource to have its accessibility improved, and another advantage of making the knowledge quicker to apply.

[0146] FIG. 8 is a diagram showing data structures for the intermediate levels of resource kb 230, that is resource kb levels one, two, three . . . eight 231, 232, 233 . . . 238. Intermediate resource kb 230 is a set of metagraphs 250. Each metagraph 250 has the information shown in Table 5.

TABLE 5
field	meaning
unique ID	A unique number assigned to each metagraph.
associated	Number of the knowledge gap rule (see Appendix A) associated with
transformation	constructing this metagraph.
number
proposed graph	Best guess of knowledge that fills the knowledge gap, written in RDF
	form.
applicable graph	The knowledge that fills the knowledge gap, being the best guess, the
	best guess confirmed by the editor, explicit knowledge supplied by the
	editor and/or implicit knowledge supplied by the editor, written in RDF
	form.
documents	The individual documents that the knowledge applied to, across a website
referencing	or document collection.
code snippet	The code for the element that is the subject of the metagraph. The code
	snippet is used by editor interface 220 to show the editor the element that
	is the subject of the editing.
object references	Reference to the equivalent of the code snippet in the application
	memory, for fast referencing and processing.

[0147] Resource description framework (RDF) is described in detail at http://www.w3.org/RDF/. RDF is a language for representing information about resources in the World Wide Web. By generalizing the concept of a “Web resource”, RDF can also be used to represent information about things that can be identified, even when they cannot be directly retrieved on the Web. RDF is intended for situations in which this information needs to be processed by applications, rather than being only displayed to people. RDF provides a common framework for expressing this information so it can be exchanged between applications without loss of meaning. RDF is based on the idea of identifying things using Web identifiers, such as Uniform Resource Identifiers, or URIs, and describing resources in terms of simple properties and property values. This enables RDF in its simplest form to represent simple statements about resources as a graph of nodes and arcs representing the resources, and their properties and values. Sometimes it is not convenient to draw graphs when discussing them, so an alternative way of writing down the statements, called triples, is also used. In the triples notation, each statement in the graph is written as a simple triple of subject, predicate, and object, in that order. Each triple corresponds to a single arc in the graph, complete with the arc's beginning and ending nodes (the subject and object of the statement).

[0148] An example of an RDF triple is as follows:

	<http://www.example.org/index.html>
	<http://www.example.org/terms/creation-date>
	“August 16, 1999”.

This RDF triple represents the knowledge that the resource index.html was created on Aug. 16, 1999.

[0149] FIG. 9 is a diagram showing the data structure for final resource knowledge base 164. Resource kb 164 comprises applicable graphs 260 from metagraphs 250. Thus, it will be appreciated that extracter 230 functions to simply copy applicable graphs 260 from metagraphs 250 into a separate file that is resource kb 164.

[0150] FIGS. 10A-10C are discussed above. FIG. 11 is a flowchart showing set-up of a user preferences file.

[0151] At step 305, user preferences setup program 156 determines whether to use a profile for the user maintained by a third party based on input from the user. In some embodiments, this step is automatically established for groups or users known to be likely to have such profiles, such as a group of students. If so, at step 325, user preferences setup program 156 gets the profile from the third party, and at step 330, user preferences setup program 156 translates the terminology of the third party profile into local terminology by mapping the third party profile's known terms to the options offered by the accessibility-improved resource and combination of scenario-specific transformations. Processing continues at step 320.

[0152] As used herein, a “pipeline” refers to a combination of scenario-specific transformations, which are different than the transformations discussed above for obtaining knowledge to fill knowledge gaps.

[0153] If a third party profile is not to be used, at step 310, user preferences setup program 156 presents a menu of different rendering features to the user; exemplary rendering features are shown in Table 6, discussed below. At step 315, user preferences setup program 156 receives the user's selections. In some embodiments a user preferences set up file is not used at all and the user is directed to an alternate rendering by activating a link on the resource.

[0154] At step 320, user preferences setup program 156 creates the user preferences file, and processing is complete.

[0155] FIG. 12 is a diagram showing a data structure for user preferences file 350, which comprises records having flag field 352 and feature field 354. Flag field 352 is set to a binary value, such as “yes” or “no”, or “0” or “1”, or “true” or “false”, indicating whether the feature in feature field 354 is or is not desired by the user. Feature field 354 indicates a rendering preferences of the user. Table 6 shows representative values for feature field 354.

TABLE 6
Field	Meaning
BW	This style is used for black and white renderings. It changes all the layout
	attributes to black and white as well as changes the links for any external
	stylesheet to connect to the black and white version of the stylesheet. In
	addition it redirects the images to go through a grayscale image filter that
	increases contrast.
disableColors	This style is also used for black and white renderings. It deletes all the
	color attributes used in the page, so that all color information will be only
	in the external stylesheet for black and white rendering.
linearizeTables	This stylesheet is used for linear rendering. It eliminates all the tables and
	creates a linear version of the page.
AccessKeys	This stylesheet is used for all the renderings except the pagemap and true
	text rendering. It deletes the text of the access keys or author preferred
	keyboard shortcuts written in the converted resource element, in order
	not to change the look of the page. An access key is a predetermined
	keyboard sequence that provides a shortcut to a destination, for example
	“alt-H” for the home page. Access keys are very important to users
	lacking a mouse, and to blind users who want to navigate other than by
	tabbing through a resource.
Basic	This stylesheet is used for all the renderings. It deletes the unnecessary or
	obsolete tags and attributes, as well as irrelevant or used converted
	resource attributes.
ClarifiedText	This stylesheet is used for simple text rendering. For each element that
	has an offshoot of the converted resource alternative tag with
	clarifiedText, it replaces the content of the original tag with the content
	of the alternative tag. Clarified text is an alternative clearer version of
	the text of a resource element.
CSS (Cascaded Style	This stylesheet is used for all the renderings. It deletes all the style and
Sheet)	layout tags as well as the attributes used in the page. All the layout and
	style information should by then be in the external stylesheet. A
	transformation is sometimes referred to as a style sheet. A cascaded style
	sheet is not a transformation.
Enhanced	The result of this rendering is insertion of a chunk of text at the start of a
Navigation	page, with the chunk comprising the hyperlinks to the headings and sub-
	headings of the page and blocks of content such as menus and sub menus,
	akin to a table of contents.
PageMap	A page map rendering creates a section at the beginning of a page with an
	interactive schematic diagram of a web resource or rerenders the
	document in pictorial form. This stylesheet creates a pagemap section at
	the beginning of the page with links to all the sections identified as
	‘pagemap blocks’, as well as to all the headings on the page. The result
	of this stylesheet is a graph of boxes, with the page title being in the
	central box, main headings being in respective boxes connected directly
	to the central box, and sub-headings being in respective boxes connected
	directly to their main heading. This format graphically depicts how the
	sections of a document are related.
TrueText	This stylesheet is used for true text renderings. It replaces all the images
	with alternative text on the page with a text box. If the image is inside a
	link, or is part of an image map, then a link will be created.
CleanUp	Removes unwanted features and tags from the converted resource that
	are not needed by the end user, such as alternative clarified text that will
	not be seen by the average user.

It will be appreciated that in some cases, multiple renderings are desired by a user, such as “bw” and “pagemap”.

[0156] Operation of the annotation wizard to create a resource knowledge base will now be discussed. In some embodiments, different interfaces and/or processes are more useful for filling knowledge gaps than the particular interface and process described below.

[0157] FIGS. 13A-13F are a flowchart showing how a resource knowledge base is created for a resource.

[0158] Turning to FIG. 13A, at step 405, annotation wizard 154 prompts the editor for session information. Session information includes the editor's username and password, a project name, the address of the resource, which standard needs to be complied with, e.g., WCAG level 1 or Section 508, and any other conversions of content that may be required, such as conversion of web content to a digital talking book. At step 410, annotation wizard 154 receives the session information provided by the editor.

[0159] At step 415, annotation wizard 154 retrieves the resource specified by the editor and creates a list of pages in the resource. For example, if the resource is a website, the list of pages sequentially lists each page in the website.

[0160] To make best use of the editor's time and the computing capacity of server 150, pages are pre-processed in parallel, and as soon as a page is pre-processed, it is available to be edited by the editor in parallel with any remaining page pre-processing. Step 420, shown in detail in FIG. 13B, indicates that pre-processing occurs in parallel with step 430, shown in detail in FIG. 13C, processing of pre-processed resources. After step 420 completes, step 430 continues to completion. The result of step 430 is resource kb level n 238, comprising metagraphs as shown in FIG. 8.

[0161] At step 435, extracter 230 of annotation wizard 154 extracts resource kb 164 from resource kb level n 238. Specifically, extracter 230 selects applicable graphs 260 from resource kb level n 238 and places them into a file to create resource kb 164.

[0162] At step 440, annotation wizard 154 provides the results of the editing session to the editor, such as code for drop-down menus enabling the user to choose an alternative rendering, and the locations of alternate renderings and combinations of scenario-specific transformations, referred to as pipelines. The code for the drop-down menus includes hyperlinks to the alternate renderings that the menu items correspond to. Processing is now complete.

[0163] Let it be assumed that accessibility engine 152 resides on a server at www.ubapps.com.

[0164] A link to an accessibility enhanced version of a website may be of the form:

• • [0165] http :/nds.ubapps.com/truetext/solutions_and_services.html
    where nds is a large client, and the accessibility enhanced site is a truetext version of the page solutions_and_services. “Truetext” is a scenario-specific transformation, also referred to as a pipeline. The resource kb is in a corresponding location available to accessibility engine 152.

[0166] Alternatively, a link may be of the form:

• • [0167] http://ubapps.com/truetext/www.nds.com/solutions_and_services/solutions_and_services.html
    if nds is a smaller client that does not require its own sub-domain and shortened URL address.

[0168] The resource owner now decides which of the drop-down menus or links should be linked to from the resource, and manually inserts the drop-down menus and/or links into the resource. Since the insertion is completely under the resource owner's control, the resource owner avoids the anxiety experienced when accessibility-enhancing software rewrites the resource.

[0169] FIG. 13B is a flowchart showing pre-processing of the pages of a resource. In short, common problems due to author sloppiness with the HTML (or XML) code are cleaned up, the page is converted to a page object, and placed on a list to await processing.

[0170] At step 450, annotation wizard 154 sets its page pointer to the first page in the list of pages created at step 415.

[0171] At step 455, annotation wizard 154 gets the page indicated by its page pointer.

[0172] At step 460, annotation wizard 154 runs a patch file on the page. The patch file includes corrections for extreme instances of bad encoding that can crash or confuse the software that creates valid XML. At step 465, annotation wizard 154 creates valid XML using third party software such as Cyberneco, available at http://www.apache.org/˜andyc/neko/doc/html/index.html, or HTML Tidy, available at http://www.w3.org/People/Raggett/tidy/. At step 470, annotation wizard 154 converts the valid XML to a page object by loading it into an XML Java object which is faster to process than an XML file. At step 475, annotation wizard 154 places the page object on either the active list or the wait list. Generally, the active list has a predetermined capacity, such as three pages. If there is room on the active list, the page object is placed on the active list; otherwise, the page object is placed on the wait list.

[0173] At step 480, annotation wizard 154 determines whether there are more pages in the list of resource pages. If so, then at step 485, annotation wizard sets its page pointer to the next page in the list of pages, and processing returns to step 455. When there are no more pages to process, at step 490, annotation wizard 154 sets an “all-pages-processed” flag, and processing is complete.

[0174] FIG. 13C is a flowchart showing processing the pre-processed resource. In short, a page is selected for page processing, then waiting pages are moved to the active list.

[0175] At step 505, annotation wizard 154 checks whether there is a page object on the active list. If so, at step 525, the page object is processed as shown in FIG. 13D, and processing returns to step 505. If there are no page objects on the active list, then at step 510, annotation wizard 154 checks whether there are page objects on the wait list. If so, then at step 520, annotation wizard 154 moves the page object to the active list and processing continues at step 525. If there are no page objects on the wait list, then at step 515, annotation wizard checks whether the “all-pages-processed” flag is set. If not, that means that some pages remain to be pre-processed, so processing returns to step 505. If the “all-pages-processed” flag is set, then processing is complete, since all pages have been pre-processed, and all pages have been processed.

[0176] FIG. 13D is a flowchart showing processing a page of a pre-processed resource. The processing depicted herein is also shown in FIG. 7.

[0177] At step 550, annotation wizard 154 gets a page object from the active list. At step 555, annotation wizard 154 sets the current resource, curr_resrc to be the just obtained page object, and sets nlev=1.

[0178] At step 560, annotation wizard 154 executes the transformations program for nlev on the current resource to create the resource kb for nlev. During the first iteration, this corresponds to executing transformations program level one 211 on page object 162 to produce resource kb level one 231 shown in FIG. 7.

[0179] At step 565, annotation wizard 154 invokes editor interface 220 to obtain knowledge for the knowledge gaps in the resource kb for nlev, shown in detail in FIG. 13E. This corresponds to providing the proposed graphs to editor interface 220 and receiving the applicable graphs from editor interface 220, shown in FIG. 7.

[0180] At step 570, annotation wizard 154 checks if there are more levels, that is, whether nlev equals the maximum number of levels of transformations programs. If so, then processing is complete. If not, at step 575, annotation wizard 154 invokes accessibility engine 152 to generate a resource at nlev, resource_nlev, shown in detail in FIG. 14B. During the first iteration, this corresponds to executing accessibility engine 152 with page object 162 and resource kb level one 231 as inputs to produce resource level one 241 shown in FIG. 7. At step 580, annotation wizard 154 sets the current resource, curr_resrc to be the just generated resource_nlev, and increments nlev. Processing continues at step 560.

[0181] FIG. 13E is a flowchart showing obtaining knowledge from an editor. Generally, wizard 154 finds the next knowledge gap in the page object and either uses an appropriate editor dialog to obtain knowledge from the editor to fill the gap or makes a “highly confident” guess as to what should fill the gap, updates proposed graph 258 with the just obtained or just guessed knowledge to create applicable graph 262, and continues until all knowledge gaps have been addressed.

[0182] At step 605, annotation wizard checks if there are any knowledge gaps. More specifically, annotation wizard checks if there are any metagraphs 250 having proposed graph 258 and lacking applicable graph 262. If not, processing is complete. If so, at step 610, annotation wizard gets the first knowledge gap.

[0183] At step 615, annotation wizard 154 checks the type of knowledge gap. It is noted that missing presentation information for the page can be a type of knowledge gap. If the knowledge gap does not require information from the editor, such as style information superseded by a style sheet, shown in Table A-1 in Appendix A, then processing continues at step 625. If the knowledge gap does require information from the editor, then at step 620, annotation wizard engages in an appropriate dialog with the editor to obtain the missing information. An example of a dialog is shown in FIG. 13F.

[0184] At step 625, annotation wizard stores the missing knowledge obtained from its own guess or from the editor. More specifically, annotation wizard 154 writes an applicable graph 262.

[0185] Tables 7-9 show examples of RDF graphs. In both of the cases discussed, information obtained by annotation wizard 154 from the editor is used to make an alternative equivalent for the given image.

[0186] Table 7 is a graph obtained by guessing that an image is a bullet (altText value=“*”, in the penultimate line), which was confirmed by the editor, so the proposed graph in Table 7 is the same as the applicable graph.

TABLE 7
<rdf:Description about=“//dt/img[count(preceding-sibling::node( )[(self::*) or (normalize-
space(.))]) &lt; 1][not(ancestor::a)][not(@alt)][not(@width &gt; 30)][not(@width &lt;
5)][not(@height &gt; 30)][not(@height &lt; 5)][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’] |
//dd/img[count(preceding-sibling::node( )[(self::*) or (normalize-space(.))]) &lt;
1][not(ancestor::a)][not(@alt)][not(@width &gt; 30)][not(@width &lt; 5)][not(@height &gt;
30)][not(@height &lt; 5)][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’] |
//li/img[count(preceding-sibling::node( )[(self::*) or (normalize-space(.))]) &lt;
1][not(ancestor::a)][not(@alt)][not(@width &gt; 30)][not(@width &lt; 5)][not(@height &gt;
30)][not(@height &lt; 5)][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’]”>
<ub:technique>1.1.5</ub:technique>
<ub:altText value=“*”/>
</rdf:Description>

[0187] Table 8 shows a proposed graph in which annotation wizard 154 guessed that the image is a spacer (altText value=“”, in the penultimate line). The editor did not agree with this guess and instead provided the information that the image represents “Services” resulting in the applicable graph shown in Table 9 (altText value=“Services”, in the penultimate line).

TABLE 8
<rdf:Description about=“//img[@width][@height][not(@width
&gt; 5)][not(@height &gt;
5)][not(@alt)][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’] |
//img[@width][@height][not(@width &gt; 5)][not(@height &gt;
5)][@alt][normalize-
space(@alt)=”][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’]”>
<ub:technique>1.1.22</ub:technique>
<ub:altText value=“”/>
</rdf:Description>

[0188]

TABLE 9
<rdf:Description about=“//img[@width][@height][not(@width
&gt; 5)][not(@height &gt;
5)][not(@alt)][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’] |
//img[@width][@height][not(@width &gt; 5)][not(@height &gt;
5)][@alt][normalize-
space(@alt)=”][not(@ismap)][not(@usemap)][normalize-
space(@src)=‘http://www.yadsarah.org.il/english/images/Gates/bb1.gif’]”>
<ub:technique>1.1.22</ub:technique>
<ub:altText value=“Services”/>
</rdf:Description>

[0189] At step 630, annotation wizard checks if there are any more knowledge gaps. If so, annotation wizard 154 gets the next gap and processing continues at step 615. If not, processing is complete.

[0190] FIG. 13F is a flowchart showing the logic for a dialog with the editor for an image without a text equivalent. It will be understood that dialogs for different situations are different, but usually follow the general format described herein.

[0191] Generally, annotation wizard 154 guesses what the missing knowledge is, then asks the editor to confirm the guess, supply implicit knowledge that is likely to exist, or supply explicit knowledge if the guess is wrong. Importantly, the editor is often prompted for information about a situation multiple times, since multiple levels of information may be conveyed in the page, and most human editors need multiple prompts to provide a full response. In other words, annotation wizard 154 is built to work with a typical human editor, and does not require an editor sensitive to various nuances of knowledge and how they might be misperceived by a disabled person.

[0192] At step 700, indicated as a box with a dashed line, annotation wizard 154 makes a guess as to the missing knowledge. In this case of an image without a text equivalent, either:

• • [0193] at step 705, the image looks important and so processing continues at step 770; in processing terms, this can be an image large enough to have readable information without text positioned on top of it, see Table A-37, or
  • [0194] at step 710, the image looks like a bullet, and so processing continues at step 730; in processing terms, this is an image of size within the range appropriate for a bullet and positioned at the beginning of a list item, see Table A-24, or
  • [0195] at step 715, the image looks like spacer or background, and so processing continues at step 740; in processing terms, this can be an image that has at least one dimension too small to contain readable information, or an image with content positioned on top thereof, see Table A-27, or
  • [0196] at step 720, the image is in a hyperlink and has text, see Table A-26, such as the following example:
    • [0197] <a href=“productlaptop.html”><img src=“laptop.gif”>see our laptops</a>so processing continues at step 745, or
  • [0198] at step 725, the image is in a hyperlink without text, see Table A-25, such as the following example:
    • [0199] <a href=“product/laptop.html”><img src=“laptop.gif”></a>so processing continues at step 750, or
  • [0200] if none of these situations exists, then processing is complete.

[0201] At step 730, annotation wizard 154 uses editor interface 220 to confirm whether the image is a bullet a bullet, at step 735, annotation wizard 154 inquires whether it is simply a bullet or whether extra information is conveyed. ff a plain bullet, processing is complete. If extra information, processing continues at step 785.

[0202] At step 740, annotation wizard 154 uses editor interface 220 to confirm whether the image is a spacer or background. If not, processing goes to step 770. If the editor confirms the image is a spacer or background, then processing is complete. This is an example of where the missing knowledge relates to presentation information.

[0203] At step 745, annotation wizard 154 uses editor interface 220 to confirm whether the text associated with the hyperlink explains the destination. If so, processing continues at step 770 to determine what the meaning of the image is. If not, processing continues at step 750.

[0204] At step 750, annotation wizard 154 guesses the role and title of the destination page by either referring to the role and title associated with the linked-to page by annotation wizard 154, or by finding the title and role of the page as it exists in the metadata of the linked-to page. At step 755, annotation wizard 154 uses editor interface 220 to confirm whether the role and title have been correctly guessed. If so, at step 765, annotation wizard 154 inquires whether there is any more information conveyed. If not, processing is complete. If so, processing continues at step 785. If the role and title have not been correctly guessed, at step 760, the editor provides the correct role and title, and processing continues at step 765.

[0205] At step 770, annotation wizard 154 uses editor interface 220 to determine whether the image conveys a message. If so, processing continues at step 785. If not, at step 775, annotation wizard 154 inquires, via editor interface 220, whether the image conveys spacing or background information. If so, processing is complete. If not, at step 780, the editor provides a description of what the image means, and processing continues at step 790.

[0206] At step 785, the editor provides the information that she or he has indicated exists. Generally, the information is provided as free-form text, without length limitation.

[0207] At step 786, the editor provides other role information such as whether this image is used as a header.

[0208] At step 790, annotation wizard 154 checks whether a summary is needed, typically, whether the information from the editor exceeds a predetermined number of characters, such as 80 characters. If not, processing is complete. If so, at step 795, annotation wizard 154 obtains, via screen interface 220, a summary from the editor, and processing is complete.

[0209] Thus, the processing depicted in FIG. 13F obtains at least the following knowledge for an image without a text equivalent:

• • [0210] the meaning of the image, if any;
  • [0211] whether the image is a bullet, and if so, any extra information associated therewith;
  • [0212] whether the image is a spacer or background; and
  • [0213] the role and title of a destination page, for a hyperlink.

[0214] Operation of the accessibility engine to provide a resource with improved accessibility will now be discussed.

[0215] FIGS. 14A-14C are a flowchart showing how a resource is converted to an accessible resource.

[0216] FIG. 14A depicts the overall processing of converting a resource to an accessible resource. In this embodiment, the resource is converted in real-time, in response to a request from a user; this approach is most practical when there are a large number of variations in how a user may prefer the final converted resource. In other embodiments, the conversion occurs during a set-up phase, and the converted resource is given directly to the user; this approach is suited to a situation where there is a small number of possibilities in how the resource may be converted.

[0217] In other embodiments, some of the conversion is performed during set-up, and the remainder is performed in response to a user request.

[0218] In still other embodiments, only one version of converted content is available and the user is another device or application such as middleware or a translation service.

[0219] At step 800, accessibility engine 152 generates a converted resource from the original resource and the resource kb. FIG. 14B shows this conversion in detail. For the case of a website, FIG. 10 depicts converting original website 162 to converted website 192 using resource kb 164 and user preferences 166.

[0220] At step 805, accessibility engine 152 gets user preferences. Typically, user preferences are provided via one or more of the following ways. First, by the drop-down menu from which accessibility engine 152 was invoked; this drop-down menu was provided to the resource owner as part of set-up session results at step 440 of FIG. 13A. Second, via a file provided by the user to accessibility engine 152, such as a cookie 115 on user pc 110. Third, via a file stored at a third party's location, such as user preferences 142 at server 130.

[0221] At step 810, accessibility engine 152 determines whether scenario-specific style sheets are needed based on the choice of rendering, where rendering specific transformations associate the correct style sheet to the page. If not, the original resource's style sheet, possibly as modified and referenced by the resource kb, for example to increase color contrast, is used and processing continues at step 820. If so, at step 815, accessibility engine 152 generates scenario-specific style sheets, as shown in FIG. 14C.

[0222] At step 820, accessibility engine 152 determines whether scenario-specific transformations should be performed. If not, the converted resource is to be delivered to the user as-is and processing continues at step 830. If so, at step 825, accessibility engine 825 applies scenario-specific transformations to the converted resource from step 805.

[0223] At step 830, the rendering is sent to the user with, if produced at step 825, a link to the scenario-specific style sheets and processing is complete. The user's browser functions to apply the scenario-specific style sheet to the rendering when presenting the rendering to the user.

[0224] FIG. 14B depicts creating a converted resource based on an original resource and a resource kb.

[0225] At step 850, accessibility engine 152 gets the original resource. At step 855, accessibility engine gets the resource kb created via annotation wizard 154, created using a different tool, or created manually.

[0226] At step 860, accessibility engine 152 uses the resource kb to alter the original resource to produce a converted resource, and processing is complete.

[0227] FIG. 14C depicts generating scenario-specific style sheets, which are a powerful mechanism for affecting the presentation of a resource. For example, a scenario specific style sheet enables a user to specify “red colors always display as bold and underlined in my black-and-white rendering”.

[0228] At step 905, accessibility engine 152 gets the original cascaded style sheet (CSS) from the converted resource; the CSS was created as part of the original resource or referenced in the converted resource. In a converted resource, CSS references were created by extracting presentation information from the original resource that had not been separated into content.

[0229] At step 910, accessibility engine 152 converts the CSS to XML so it can be parsed using the parser developed for parsing the XML knowledge gap rules.

[0230] At step 915, accessibility engine 152 gets the next criterion from scenario-specific style sheet rules 168. An example of a criterion is “turn all red text into bold and underlined text”, and is expressed as a pattern to be matched in the resource's code.

[0231] An example of a scenario-specific style sheet rule is shown in Table 10. This rule converts colors to grayscale.

TABLE 10
<transformation>
<number>3</number>
<logic>Convert colours into either black or white. This is based on the
background colour</logic>
<usedFor>Black-white renderings.</usedFor>
<typeReference>2</typeReference>
<case>
<description>Matched on a background-color attribute</description>
<if path=“//” type=“attribute” name=“background-color”/>
</case>
<case>
<description>There is a css:color attribute but no background-color
in the same node</description>
<if path=“//” type=“attribute” name=“color”/>
</case>
<action>
<description>Change the background colour to the base colour and the
font colour to the inverse of that</description>
<then action=“changeColourValue”/>
<then action=“changeBackgroundValue”/>
</action>
</transformation>

[0232] Exemplary rules in scenario-specific style sheet rules 168 are:

• • [0233] Convert the base font size in the body element tag to em so that it is relative;
  • [0234] Convert any font-size that is not relative (em) to relative. Either to the base font-size that has already been set or to the default which is lem;
  • [0235] Convert colours into either black or white. This is based on the background colour;
  • [0236] Convert bolder or underlined text to louder;
  • [0237] Convert font-weight bold to stress: 90; pitch-range: 80;
  • [0238] Take a font-size and add richness tag depending on what the relative font-size is.

[0239] At step 920, accessibility engine 152 determines whether instances of this criterion exist in the resource. If so, at step 925, the instance is changed with the appropriate rule, and at step 930, accessibility engine 152 determines if there are any more instances. If so, processing returns to step 925. If not, processing continues at step 935. If no instances of this criterion exist, then processing returns to step 915.

[0240] At step 935, accessibility engine 152 determines if there are any more criteria. If so, processing returns to step 915. If not, at step 940, accessibility engine 152 converts the XML code back to a CSS, and processing is complete. Using a mapping of CSS to XML enables use of XML type identifiers such as URIs or Xpointers, to describe rules and relationships in a non-XML based platform.

[0241] Although the present invention has been described with regard to websites, it will be appreciated that it is also useful with regard to any electronic resource. That is, the present invention is adapted for converting documents such as graphics animations or presentations to be accessible by visually impaired persons.

[0242] Usage examples are now discussed.

[0243] Consider an indefinite object, such as “it”, appearing in a sentence. In context, the average user will have little difficulty determining the definite object referred to, but the cognitively challenged user may find understanding the implied meaning of the word “it” difficult. If the sentence or phrase containing the word “it” is cut and pasted into another document, the meaning of the word “it” will be lost. If however, the definite object referred to by the word “it” is determined and stored in the knowledge base, a link to the implied meaning of the word may be transported with the phrase between documents maintaining accessibility to the meaning.

[0244] Another application that illustrates the advantages of this approach is translation. Translation tools that automatically translate between languages are known. However it will be appreciated that sentence construction, word order and basic assumptions regarding the connection between words vary between languages. It is well known that most written material ‘loses something in the translation’, even when the translation is performed manually by a bilingual individual. Where translation is performed by a machine, for the translation to be comprehensible, the machine working with the text must be very sophisticated; unfortunately, state of the art automatic translators are inadequate. By including appropriate supplementary material, a knowledge base as described above enables better translations to be performed.

[0245] Another example of knowledge, implicit in a page, is the role of different interactive elements. If we know that a page hyperlink has the role of taking the user to the site's home page, and we store that piece of knowledge in the knowledge base, then that knowledge can be used to create enhanced accessibility in different ways in many different scenarios. For example, the home page hyperlink and all similar elements can be rendered for a user, across all pages, with an keyboard access alternative of “alt+h”. However, if we consider a different user, such as a native Russian speaker, and that ‘home’ in Russian, is “dom”, it may be preferable to use “alt+d” to consistently provide keyboard access for links to the home page, across any site. Similarly, a cognitively challenged user may prefer all links to home pages to be represented by an icon of a home, perhaps with the site name displayed underneath. This example illustrates the advantages of comprehending the meaning implicit in a link, rather than merely offering a particular substitute.

[0246] Another example is as follows. Let it be assumed that a screen displays the sentence:

• • [0247] “RDF is written in triples.”
    A user can select a zoom feature, which causes accessibility engine 152 to locate an explanation of terms, and revise the displayed sentence to be as follows:
  • [0248] “RDF (Resource Description Framework) is written in triples (triples describe the relationship between two items).”
    This feature allows accessibility engine 152 to expand the content for the user so that the user does not have to remember each term. This is especially useful for memory related disorders where the user is able to understand the concepts but unable to remember terms. It can be appreciated how much easer this interface would be then just providing a glossary look up, or requiring the user to go back and forth looking up the terms via a dictionary or web search. The same type of zoom could be applied to many scenarios such as diagrams and equations, where sections or terms can be expanded into more detail, explanations or dependent sub equations. Legal documents can zoom into defined terms and first usages. Legal document and technical documents, using removing ambiguity and zoom could be made clearer and more comprehensible to many more people.

[0249] Although an illustrative embodiment of the present invention, and various modifications thereof, have been described in detail herein with reference to the accompanying drawings, it is to be understood that the invention is not limited to this precise embodiment and the described modifications, and that various changes and further modifications may be effected therein by one skilled in the art without departing from the scope or spirit of the invention as defined in the appended claims.

[0250] The following terms are used herein:

CSS	cascaded style sheet
SWAP	an embodiment of the present invention that has been reduced
	to practice
WCAG	Web Content Authoring Guide

[0251]

TABLE A-1
Field	Description
Dependency	A-1
Transformation	3.3.11
Number
Knowledge gap	Associating content and style
Logic	Body tag with margin attributes
Used for	Take out all the presentational attributes from the body tag and put them into
	CSS.
About	Presentational information should be separated from other content. Where
	they have been mixed up it is useful to separate them. When we do we create
	a knowledge base reference linking styled content in the page to style
	information in a class in a style sheet. We need to build the knowledge base
	relationship and, when we do not have it already add style information to a
	CSS that acts as a resource. In this case we know we need to do this because
	of the occurrence body tag with margin attributes
Best guess	Create CSS class based on presentational information (unless one exists
	already).
	Create association between content and CSS
	This is normally automatic
	User can edit if they choose
Editor supplied	None
knowledge
Example case	Case 1: if there are explorer margin attributes in the body tag, we need to
(HTML	transfer them all to CSS
description)	Case2: if there are explorer margin attributes in the body tag, we need to
	transfer them all to CSS
Example use,	writeToCSSFile creates a new class with attribute of current node deletes all
WCAG	these attributes from the node and adds to the node attribute class with the
description	value of the new class name created All presentational attributes are deleted
	in SWAPML rendering SWAPCSS.xsl. NOTE: there may be a problem
	because this transformation now will not have an rdf statement

[0252]

TABLE A-2
Field	Description
Dependency	A-1
Transformation	3.3.12
Number
Knowledge gap	Associating content and style
Logic	body tag with margin attributes
Used for	take out all the presentational attributes from the body tag and put them to
	css.
About	Presentational information should be separated from other content. Where
	they have been mixed up it is useful to separate them. When we do we create
	a knowledge base reference linking styled content in the page to style
	information in a class in a style sheet. We need to build the knowledge base
	relationship and, when we do not have it already add style information to a
	css that acts as a resource. In this case we know we need to do this because
	of the occurrence body tag with margin attributes
Best guess	Create CSS class based presentational information (unless one exists already)
	Create association between content and CSS
	This is normally automatic
	User can edit if they choose
Editor supplied	None
knowledge
Example case	Case 1: if there are netscape margin attributes in the body tag and no explorer
(HTML	attributes, we need to transfer them all to css
description)	Case2: if there are netscape margin attributes in the body tag, we need to
	transfer them all to css
Example use,	writeToCSSFile creates a new class with attribute of current node deletes all
WCAG	these attributes from the node and adds to the node attribute class with the
description	value of the new class name created All presentational attributes are deleted
	in SWAPML rendering SWAPCSS.xsl. NOTE: there may be a problem
	because this transformation now will not have an rdf statement

[0253]

TABLE A-3
Field	Description
Dependency	A-2
Transformation	15.1.1
Number
Knowledge gap	Role of content
Logic	for every page, identify its type (front matter, bodymatter, rearmatter)
Used for	convert to daisy book
About	Role of content is important piece of knowledge about a resource.
	For every page, identify its type and role (front matter, bodymatter,
	rearmatter)
	Note on roles:
	Adding new roles and classes - Extendibility:
	SWAP Roles and Types are defined in RDF and OWL classes. That makes
	them inherently extendable and new classes and types of roles can be derived
	from old existing roles. The relationship between new classes and types and
	preset types can also be defined in a programmatically understandable way.
	This implies web communities can have their own, derived, content types.
	For example IMPS and educational content typically use:
	tutor, help, glossary, back, up, next
	This implies that, if you customers feel an element usage is not fully
	described in the given role libraries then they can create a new role. This is
	especially important to niche web communities such as bloggers, who may
	expect different types of content on a web site.
	The invention can also come with an easy interface to create new types of
	content types. New sets of content types can be down loaded and used by r
	customers
	What happens at the user agent does not know about a new type?
	If the user agent is only familiar with the base classes, then it can handle a
	derived class the same way it would handle its parent. However tools geared
	to a certain target audience (such as educational) would have special features
	for a derived schema created by that community.
	Popular downloaded set of types are also supported by SWAP interpretation
	engine
Best guess	If the CSS classes of the page (body class) has a role associated then we will
	guess that role
	Guess depends on where the page is in the hierarchy of the site/content
	(contact us is more likely to be a link off the home page, table of content
	comes at the front of a book, etc.)
	Also the content itself helps us guess the role (lots of links - may be a site
	map)
	And the content of a page header, title etc helps us guess the role.
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	Case 1: get a container with page class
(HTML
description)
Example use,	ask the user for the role attribute
WCAG
description

[0254]

TABLE A-4
Field	Description
Dependency	A-3
Transformation	13.2.1
Number
Knowledge gap	Description of page
Logic	meta tag with description name is missing in the text
Used for	insert essential metadata to the document
About	meta tag with description name is missing in the text
	(note this may have been added when
	another page was run in the wizard and
	referenced this page. In that case we will
	already know this information, - in which case we
	will not run this transformation, as it is added
	at level 1 transformations)
Best guess	insert essential metadata to the document
	from first paragraph of the main
	content
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	Case1: meta tag with name = description
(HTML	is not found
description)
Example use,	need to add meta tag with
WCAG	name = description
description

[0255]

TABLE A-5
Field	Description
Dependency	A-3
Transformation	13.2.2
Number
Knowledge gap	Title and role of a page
Logic	title tag is missing in the text
Used for	insert title for the document
About	Meta data for Role and Title for a page is
	important to help users find the
	page that they need and navigate a site \.
	title tag is missing in the text
	Title tag was used by another page
	role is not known
	(note this may have been added when
	another page was run in the wizard and
	referenced this page. In that case we
	will already know this information, - in
	which case we will not run this transformation,
	as it is added at level 1 transformations)
Best guess	Role may be associated with CSS of the
	body of a page, or be associated
	from the page.
	Title can be guessed as being equal to
	the Role Otherwise,
	insert title for the document - if there is a first header
	in the content then that may be there title. Role can be
	guessed from title keywords or may have been
	provided when this page was referenced by a page we
	already processed All titles should be
	unique (pages on a site should not have the
	same title, but can have the same role)
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	Case1: title tag is not found in the document
(HTML	Case2: role is not known
description)	Case 3: title has been used by a page
	that is already processed.
Example use,	need to add title tag
WCAG
description

[0256]

	TABLE A-6
	Field	Description
	Dependency	A-3
	Transformation	13.3.1
	Number
	Knowledge gap	n/a
	Logic	add a class for pagemap in CSS file
	Used for	create a page map for a document
	About	add a class for pagemap sections in CSS file
	Best guess	automatic
	Editor supplied	None
	knowledge
	Example case	if there is no tag with id or class “pagemap”,
	(HTML	add one as a first node of the body
	description)
	Example use,	add a CSS class for pagemap
	WCAG
	description

[0257]

TABLE A-7
Field	Description
Dependency	A-4
Transformation	9.5.01
Number
Knowledge gap	Author preferred keyboard access
Logic	pagemap anchor tag - need to provide an
	accesskey for appropriate pagemap link
Used for	provide keyboard access to pagemap links
About	provide keyboard access (so users who cannot
	use a mouse can easily access links)
	pagemap anchor tag - need to provide an accesskey
	for appropriate pagemap link
Best guess	Suggest pagemap standard defaults
Editor supplied	None
knowledge
Example case	pagemap anchor tag will need an accesskey
(HTML
description)
Example use,	need to add accesskey
WCAG
description

[0258]

TABLE A-8
Field	Description
Dependency	A-4
Transformation	9.5.6
Number
Knowledge gap	Author preferred keyboard access
	Relative importance of links
Logic	a tag without accesskey attribute
Used for	provide keyboard access to important links
About	a tag without accesskey attribute
Best guess	If the role is known then suggest using
	role defaults access keys and not
	overriding. Recommend assigning
	the next available page accesskey for
	keyboard access to important links without
	roles by marking their relative importance
	Important tags will get higher priority
	to receive keys allocated first.
	Allow editor to override, automatic
	allocation by assigning specific
	access keys
Editor supplied	Confirm edit or change the best guess
knowledge	Allow editor to override, automatic
	allocation by assigning specific access keys
Example case	Case1: a tag with no accesskey attribute,
(HTML	with href and content
description)	Case2: a tag with no accesskey attribute,
	with href and tag inside
Example use,	need to add accesskey with next value
WCAG	available We make a suggestion for the accesskey
description	value, using java extension function that
	each time returns the next letter.

[0259]

TABLE A-9
Field	Description
Dependency	A-4
Transformation	9.5.7
Number
Knowledge gap	Author preferred keyboard access
	Relative importance of links
Logic	area tag without accesskey attribute
Used for	provide keyboard access to important links
About	area tag without accesskey attribute
Best guess	If the role is known then suggest using role
	defaults access keys and not overriding.
	Recommend assigning the next available
	page accesskey for keyboard access to
	important links without roles by marking
	their relative importance
	Important tags will get higher priority to receive
	keys allocated first. Allow editor to override,
	automatic allocation by assigning specific
	access keys
Editor supplied	Confirm edit or change the best guess. Allow
knowledge	editor to override, automatic allocation
	by assigning specific access keys
Example case	area tag with no accesskey attribute
(HTML
description)
Example use,	need to add accesskey with next
WCAG	value available
description

[0260]

TABLE A-10
Field	Description
Dependency	B-3
Transformation	1.1.13
Number
Knowledge gap	Function, role and equivalency of applet
Logic	applet with no or empty alt attribute or empty content
Used for	applet
About	Applets are little programs running on a page.
	If they are without a text equivalent then
	knowledge and/or functionality is missing
Best guess	Parser in screen should help build Knowledge
	to include functional equivalent.
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	Case1: applet with no alt attribute
(HTML	Case2: applet with empty alt attribute
description)	Case3: applet with empty content
Example use,	Adds an alt attribute. Add a text
WCAG	content for applet
description

[0261]

TABLE A-11
Field	Description
Dependency	B-4
Transformation	1.1.3
Number
Knowledge gap	Possible Function, role and equivalency of applet
Logic	applet with non-empty alt - should verify
	that the alt is sufficient
Used for	applet
About	Applets are little programs running on a page.
	If they have a text equivalent then we
	should verify and test that this knowledge is full
Best guess	Confirm with user
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	applet whose alt attribute is not empty
(HTML
description)
Example use,	Verify that alt attribute is sufficient. If not,
WCAG	edit text content
description

[0262]

	TABLE A-12
	Field	Description
	Dependency	C-2
	Transformation	11.2.7
	Number
	Knowledge gap	Map to XHTML
	Logic	deprecated embed tag is used in the document
	Used for	replace embed tag with object tag
	About	deprecated embed tag is used in the document
	Best guess	map embed tag to object and attributes
	Editor supplied	None
	knowledge
	Example case	embed tag is found in the document
	(HTML
	description)
	Example use,	replace deprecated embed tag with object
	WCAG
	description

[0263]

TABLE A-13
Field	Description
Dependency	C-3
Transformation	1.1.14
Number
Knowledge gap	Possible Function, role and equivalency
Logic	a programmatic object with no text content (object
	doesn't have any child object nodes)
Used for	object
About	A programmatic object such as FLASH that is
	contained in the page - full interaction mapping
	is needed in wizard for full text equivalent (object
	doesn't have any child object nodes)
Best guess	Parser called by screen and pre known
	mappings should help build knowledge to
	include functional equivalent in html or x-form.
Editor supplied	Confirm edit or change the best guess add
knowledge	new information where no guess exists
Example case	programmatic object with no text content that doesn't
(HTML	have any child object nodes (if there are few objects
description)	nested in one another, description is provided
	in content of the innermost one) if object has
	embed tag as a child, required text content is
	provided in content of embed tag, and not in
	object tag - transformation 1.1.23
Example use,	put UBcodeAdd to the object content and
WCAG	also add a long description
description

[0264]

TABLE A-14
Field	Description
Dependency	C-6
Transformation	1.1.24
Number
Knowledge gap	Knowledge if this element contains flicker
	Possible Function, role and equivalency
Logic	for any object, verify its alt attribute is provided, and
	check if it flickers
Used for	confirm alt attributes and deal with flickering
About	for any object that seems to have a text equivalent we
	need to verify that all information and text is captured,
	and check if it flickers and moving
	Check that this is not used as a header or other role
Best guess	This is not a header and does not have structural
	importance
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	for any programmatic object, verify alt and check
(HTML	for flickering
description)
Example use,	Remove flickering objects with an equivalent that does
WCAG	not flicker for photosensitive epilepsy
description

[0265]

TABLE A-15
Field	Description
Dependency	D-3
Transformation	12.1.1
Number
Knowledge gap	frame title, role, content type and description
Logic	a frame without title and longdesc attributes
Used for	frame
About	Frames are sections of a screen that function
	independently.
	They can be hard to navigate with assistive technology.
	This case looks at frames without titles and descriptions
Best guess	Title/role of page referenced by frame. Meta data
	summary or description of referenced page
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	frame without both title and longdesc
(HTML
description)
Example use,	provide title and write longdesc for the frame
WCAG
description

[0266]

TABLE A-16
Field	Description
Dependency	D-3
Transformation	12.1.2
Number
Knowledge gap	frame title, role, content type and description
Logic	a frame without title attribute but with longdesc attribute
Used for	frame
About	a frame without title attribute but with longdesc attribute
Best guess	Title/role of page referenced by frame
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	frame without title attribute but with longdesc attribute
(HTML
description)
Example use,	provide title for the frame
WCAG
description

[0267]

TABLE A-17
Field	Description
Dependency	D-3
Transformation	12.1.3
Number
Knowledge gap	frame description
Logic	a frame without longdesc but with title attribute
Used for	frame
About	a frame without longdesc but with title attribute
Best guess	Meta data summary or description of referenced page
	or ask user to provide additional information when a title
	does not suffice to adequately convey the function
	or role of a frame
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	frame without longdesc but with title attribute
(HTML
description)
Example use,	write longdesc for the frame
WCAG
description

[0268]

TABLE A-18
Field	Description
Dependency	D-4
Transformation	1.1.27
Number
Knowledge gap	frame description
Logic	frame tag with no longdesc attribute
Used for	provide additional information when a title does not
	suffice to adequately convey the function or role
	of a frame
About	frames are when sections of a page act independently.
	Sometimes a long description of the frame is
	useful but missing attribute
Best guess	Meta data summary or description of referenced page
	or ask user to provide additional information when a
	title does not suffice to adequately convey the
	function or role of a frame
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: longdesc attribute is missing in frame
(HTML	Case2: longdesc attribute is empty in frame
description)
Example use,	write longdesc for the frame
WCAG
description

[0269]

TABLE A-19
Field	Description
Dependency	E-3
Transformation	1.2.1
Number
Knowledge gap	Role of map,
	equivalent of map
	Titles/role of each linked to region, and link address
	information, summary
Logic	img tag is used to insert server-side image map
Used for	provide an alternative list of links after it and indicate
	the existence and location of the alternative list
About	An image map sits on the server and is harder for
	assistive technology to understand. we need more
	knowledge of all the destination links or indicate
	the existence and location of the alternative list
Best guess	Titles/role of each linked to region, and link address
	information
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	img is used to present server-side map
(HTML
description)
Example use,	need to provide a paragraph with alternative list of links
WCAG	after the a tag
description

[0270]

TABLE A-20
Field	Description
Dependency	E-3
Transformation	1.5.1
Number
Knowledge gap	knowledge missing includes provide redundant text in
	the links for each active region of a client-side
	image map
Logic	“A” tag is used in client-side image map instead of area
	but no content for links is provided
Used for	provide redundant text in the links for each active region
	of a client-side image map
About	a link tag is used in client-side image map instead of
	area but no content for links is provided
Best guess	Titles/role of each linked to region
	(if roles are not yet allocated to linked to pages we
	can set them now)
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	no link text is provided when a tag is used inside the
(HTML	client-side image map
description)
Example use,	need to provide textual link
WCAG
description

[0271]

TABLE A-21
Field	Description
Dependency	F-3
Transformation	2.1.1
Number
Knowledge gap	What is meant by text
Logic	check that color names do not appear together with words
	like select/if/choose/click
Used for	check that color names are not used to convey
	information that is not presented in other ways
About	If colors are used to give instructions, then other way
	of capturing that knowledge needs to be provided.
	For example saying “select the blue button” is not
	helpful for or clear for color blind people. This is
	likely to happen when instructions contain color names
	together with words like select/if/choose/click etc
Best guess	Need to bind color (eg blue) with control
	(eg. The blue button)
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	for each text content, check if it contains a sentence, that
(HTML	contains both a common color name and a selection word
description)
Example use,	Supply clarified text.
WCAG
description

[0272]

TABLE A-22
Field	Description
Dependency	H-1
Transformation	1.1.18
Number
Knowledge gap	Role of button may need to be FULLY captured
Logic	Input as an image with existing alt tag - need to verify it
Used for	text equivalency can be used as an alt attribute or as the
	text on a button in a true text rendering. The text
	equivalent of every single button on the page has
	to be unique.
About	Form control input has a text equivalent but we need
	to check that it is good enough. The text equivalent
	of every single button on the page should be
	unique. And text should be clear (not like ‘click hear’..)
Best guess	Role of a control can often be guessed by its tag, type
	and form in which it belongs
Editor supplied	Confirm edit or change the best guess
knowledge
Example case	input (inside a form) is a graphic with alt attribute
(HTML
description)
Example use,	verify the uniqueness of alt attribute - replace
WCAG	with a more specific and useful alt when necessary
description

[0273]

TABLE A-23
Field	Description
Dependency	H-1
Transformation	1.1.19
Number
Knowledge gap	more knowledge may need to be captured as to role
	of button
Logic	Button contains an image with alt attribute
Used for	text equivalency can be used as an alt attribute or as
	the text on a button in a true text rendering. Image button
	alt text should be unique for every single button
	on the page.
About	What happens when we click a button needs to be fully
	clear unique.
Best guess	Add role/name of form to button
	If form role is not known it can be provided now
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	button (inside a form) is a graphic and the alt attribute
(HTML	exists for the descendant img tag set
description)
Example use,	set the alt attribute to “go [form name]”
WCAG
description

[0274]

TABLE A-24
Field	Description
Dependency	H-2
Transformation	1.1.5
Number
Knowledge gap	Role of image
Logic	images at the beginning of definition lists created with
	DL, DT, and DD and unordered lists - LI, where there
	is no alt attribute for text equivalency
Used for	dt, dd and li with image as a first child
About	Image without text equivalent - Small images at the
	beginning of list items, are probably bullets
Best guess	this image is a bullet? Is other additional information
	in this image?
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: img inside a dt that has no alt attribute, is small,
(HTML	not inside a link and in the first position
description)	Case2: img inside a dd that has no alt attribute, not inside
	a link and in the first position
	Case3: img inside an li that has no alt attribute, not inside
	a link and in the first position
Example use,	add alt attribute
WCAG
description

[0275]

TABLE A-25
Field	Description
Dependency	H-2
Transformation	1.1.6
Number
Knowledge gap	The role of the image,
Logic	when an image is used as the content of a link, but no alt
	attribute and no link text are provided
	for text equivalency
Used for	links
About	Image without text equivalent - when an image is used
	inside a link, but no text equivalents the text equivalent
	should probably be were the link goes to
Best guess	this image is used as a link to page of
	known title and/or role
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: an image is used as the content of a link and
(HTML	neither link text nor alt attribute is provided
description)	Case2: an image is used as the content of a
	link and alt attribute is empty
Example use,	Eg.
WCAG	add alt attribute with value go to title of the page
description

[0276]

TABLE A-26
Field	Description
Dependency	H-2
Transformation	1.1.7
Number
Knowledge gap	Role of image - is it decoration only?
Logic	when an image is used as the content of a link, there is a
	link text, but no alt attribute is provided
Used for	links
About	Image without text equivalent - when an image is used
	inside a link, but there is linked text the text
	equivalent may not be necessary
Best guess	Role of image - is it decoration only?
	knowledge may be needed . . .
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: an image is used as the content of a link there is a
(HTML	link text but alt attribute is not provided
description)	Case2: an image is used as the content of a link there is a
	link text but alt attribute is not provided
Example use,	add alt attribute with value “ “(alt should be blank)
WCAG
description

[0277]

TABLE A-27
Field	Description
Dependency	H-2
Transformation	1.1.22
Number
Knowledge gap	Role of image
Logic	spacer image - relatively small one
Used for	need an empty alt attribute - if has one, check that the
	value of alt is blank, else set the alt to be blank
About	Image without text equivalent that has very small
	dimensions may be a spacer image used for
	indentions, or background. For example, an image
	with a low width (smaller the a word) or
	height shorter then a letter.
Best guess	this image has no knowledge behind it, Its
	role is formatting only
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: “small” img without an alt attribute
(HTML	Case2: “small” img with empty attribute
description)
Example use,	set alt attribute to be (blank)
WCAG
description

[0278]

TABLE A-28
Field	Description
Dependency	H-2
Transformation	1.1.10
Number
Knowledge gap	Equivalents and function and role of a map area
Logic	a client side image map with no or empty alt in area tag
Used for	map/area
About	An Image map is a picture with different clickable
	regions. If there is no text equivalent, knowledge capture
	should probably be about were each regain
	links to (such as the title or role of a destination page
Best guess	capture the name/title role of each linked to resource.
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: an area inside a map with empty alt attribute
(HTML	Case2: an area inside a map with no alt attribute
description)
Example use,	Add alt attribute with go to name/title of the page
WCAG
description

[0279]

TABLE A-29
Field	Description
Dependency	H-2
Transformation	1.1.12
Number
Knowledge gap	Role of the image ma
Logic	for any image used for client or server-side image
	map, need to provide an alt
Used for	provide alt attribute for image used for a map
About	With an image map, knowledge is needed about the
	whole map - Eg, a summary on how to use the map and
	what content is contained, expected, and other
	information such as how clickable areas connect
Best guess	Ask editor
Editor supplied	General knowledge about the whole map
knowledge
Example case	Case1: for image used as a server-side image map,
(HTML	provide an alt attribute, if doesn't have one
description)	Case2: for image used as a client-side image map,
	provide an alt attribute, if doesn't have one
Example use,	add alt attribute
WCAG
description

[0280]

TABLE A-30
Field	Description
Dependency	H-2
Transformation	1.1.16
Number
Knowledge gap	Role of control
Logic	Input as an image where there is no alt attribute
	for text equivalency
Used for	text equivalency can be used as an alt attribute or as the
	text on a button in a true text rendering
About	Form controls sometimes need text equivalent (such as
	an image used as a input control)
Best guess	Often we can get this information from the control type
	and form name and action
	If we do not have the form role, we can get it now
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: input (inside a form) is a graphical
(HTML	button without alt attribute
description)	Case2: input (inside a form) is a graphic
	with empty alt attribute
Example use,	set the alt attribute to “go [form name]”
WCAG
description

[0281]

TABLE A-31
Field	Description
Dependency	H-2
Transformation	1.1.17
Number
Knowledge gap	Role of control
Logic	Button contains an image where there is no
	alt attribute for text equivalency
Used for	text equivalency can be used as an alt attribute or as the
	text on a button in a true text rendering
About	Form controls sometimes need text equivalent In this
	case we have a button that does not have text
	explaining its role (may use an image)
Best guess	we can get this information from the control
	type and form name and action
Editor supplied	Confirm edit or change the best guess add new
knowledge	information where no guess exists
Example case	Case1: button (inside a form) is a graphic and there
(HTML	is no alt attribute for the descendant img tag set
description)	Case2: button (inside a form) is a graphic and there is
	an empty alt attribute for the descendant img tag
Example use,	set the alt attribute to “go [form name]”
WCAG
description

[0282]

TABLE A-32
Field	Description
Dependency	H-2
Transformation	1.1.30
Number
Knowledge gap	Full description and information in an image
Logic	img tag that needs additional information to be provided
	and doesn't have a longdesc attribute img needs
	additional information, if it is “big” enough.
	Also, need to provide description link for user agents
	that don't support longdesc
Used for	provide additional information when a short text
	equivalent does not suffice to adequately convey
	the function or role of an image
About	img may need additional information This is typical of
	large images, provide additional information when a
	short text equivalent does not suffice to adequately
	convey the function or role of an image and all
	text that may be on it
Best guess
Editor supplied	knowledge is needed about the long description
knowledge
Example case	Case1: longdesc is missing in “big” img and d-link is
(HTML	missing and the img is not inside a link
description)	Case2: longdesc is missing in “big” img and d-link is
	missing and the img is not inside a link
	Case3: longdesc is missing in “big” img and d-link is
	missing and the img is inside a link
	Case4: longdesc is missing in “big” img and d-link is
	missing and the img is inside a link
Example use,	Add d link and long desc attribute to img
WCAG
description

[0283]

	TABLE A-33
	Field	Description
	Dependency	H-2
	Transformation	15.1.2
	Number
	Knowledge gap
	Logic	For every tag with a style, move the style to css
		and get a class attribute
	Used for	daisy book
	About	For every tag with a style, move the style to css
		and get a class attribute
	Best guess	daisy book
	Editor supplied	None
	knowledge
	Example case	Move style attribute to css, and construct a
	(HTML	class attribute. The positioning should stay in the
	description)	style (position, top and left properties), the
		order of the properties should be insignificant.
	Example use,	create class attribute and maybe leave style
	WCAG	attribute with positioning predicates
	description

[0284]

TABLE A-34
Field	Description
Dependency	H-2
Transformation	3.4.1
Number
Knowledge gap	Map to relative units
Logic	For each attribute that specifies units of measure, check that relative rather
	than absolute units are used, e.g. that em and % are used instead of pt, cm
	etc.
Used for	check for use of relative rather than absolute units of measure
About	Relative units are ones that expand easy at the user request which is good for
	enlargements. For each attribute that specifies units of measure, check that
	relative rather than absolute units are used, e.g. that em and % are used
	instead of pt, cm etc. When non expandable units are sued we need to map to
	relative ones
Best guess	Automatically convert absolute to relative equivalent
	Algorithm takes cascading affect into account and them maps the “total” size
	relative to the base size, and maps to equivalent non relative size
Editor supplied	None
knowledge
Example case	Case1: if there is an attribute that contains measure units, it should contain
(HTML	em or %, otherwise it uses absolute units. The only exception is img tag
description)	where absolute measure units are allowed.
	Case2: if there is a style attribute that contains css property that needs
	measure unit, it should contain em or %, otherwise it uses absolute units.
	Case3: if there is a style tag that contains css property that needs measure
	unit, it should contain em or %, otherwise it uses absolute units.
Example use,	Put relative units and sizes in HTML
WCAG
description

[0285]

TABLE A-35
Field	Description
Dependency	H-2
Transformation	3.4.2
Number
Knowledge gap	Map to relative units
Logic	For each attribute that specifies units of measure, check that relative rather
	than absolute units are used, e.g. that em and % are used instead of pt, cm
	etc.
Used for	check for use of relative rather than absolute units of measure
About	When non expandable units are sued we need to map to relative ones. For
	each attribute that specifies units of measure, check that relative rather than
	absolute units are used, e.g. that em and % are used instead of pt, cm etc.
Best guess	Automatically convert absolute to relative equivalent
	Algorithm takes cascading affect into account and them maps the “total” size
	relative to the base size, and maps to equivalent non relative size
Editor supplied	None
knowledge
Example case	Case1: if there is a style attribute that contains css property that needs
(HTML	measure unit, it should contain em or %, otherwise it uses absolute units.
description)	Case2: if there is a style tag that contains css property that needs measure
	unit, it should contain em or %, otherwise it uses absolute units.
Example use,	Put relative units and sizes in HTML
WCAG
description

[0286]

TABLE A-36
Field	Description
Dependency	H-2
Transformation	3.4.3
Number
Knowledge gap	Map to relative units
Logic	If font tag has size attribute, it should contain + or −
	signs, e.g. it shouldn't use absolute measure units
Used for	check for use of relative rather than absolute units of
	measure
About	When non expandable units are sued we need to map to
	relative ones. If font tag has size attribute, it should
	contain + or − signs, e.g. it shouldn't use absolute
	measure units
Best guess	Automatically convert absolute to relative equivalent
	Algorithm takes cascading affect into account and them
	maps the “total” size relative to the base size,
	and maps to equivalent non relative size
Editor supplied	None
knowledge
Example case	if there is a size attribute in font tag, it should
(HTML	contain + or −, otherwise it uses absolute units.
description)
Example use,	Put relative units and sizes in HTML
WCAG
description

[0287]

TABLE A-37
Field	Description
Dependency	H-3
Transformation	1.1.8
Number
Knowledge gap	Meaning, information and description of an image
Logic	an image with no alt attribute for text equivalency.
	An image is not a bullet and it's not a spacer image
	and not in a link
Used for	Images
About	Image without text equivalent - not a known case
	but large enough width and height to have content,
	although it is not embedded in a link or positioned
	as a button or bullet
Best guess	if the image has meaning or information, then that needs
	to be known. If not then the image should probably
	be described
Editor supplied	Editor need to supply knowledge
knowledge
Example case	Case1: an image without an alt and is not a spacer image
(HTML	and not a bullet (from dependencies)
description)	Case2: an image with empty alt and is not a spacer image
	and not a bullet (from dependencies)
Example use,	add alt attribute with value
WCAG
description

[0288]

TABLE A-38
Field	Description
Dependency	H-3
Transformation	3.5.3
Number
Knowledge gap	Structural role of the paragraphs
Logic	header tag used for formatting
Used for	Replace header tag used for formatting with p tag using bold (options may be
	p with stylesheet or blockquote tag).
About	Header tags are sometimes used for formatting text that in fact is not a
	header. In other words - tags have been used incorrectly and provide
	misinformation to assistive technologies. The true structural information
	needs to be captured. Headers the are relatively long, is an example were this
	may occur
Best guess	These are paragraphs and not headers
Editor supplied	Confirm, edit or change the best guess add new information where necessary
knowledge
Example case	h1 tag is used for formatting - content is too long - more than 100 characters
(HTML	(may be changed
description)
Example use,	replace the header tag with p and class name we need to create css classes
WCAG	h1-h6 for each heading level
description

[0289]

TABLE A-39
Field	Description
Dependency	H-3
Transformation	3.5.4
Number
Knowledge gap	Structural role of the paragraphs
Logic	header tag used for formatting
Used for	replace header tag used for formatting with p tag using
	bold (options may be p with stylesheet or blockquote tag)
About	other catches for header tag used for formatting
Best guess	These are paragraphs and not headers
Editor supplied	Confirm, edit or change the best guess add new
knowledge	information where necessary
Example case	h2 tag is used for formatting - content is too
(HTML	long - more than 100 characters(may be changed)
description)
Example use,	replace the header tag with p and class name we need
WCAG	to create css classes h1-h6 for each heading level
description

[0290]

TABLE A-40
Field	Description
Dependency	H-3
Transformation	3.5.5
Number
Knowledge gap	Structural role of the paragraphs
Logic	header tag used for formatting
Used for	replace header tag used for formatting with p tag using
	bold (options may be p with stylesheet or blockquote tag)
About	other catches for header tag used for formatting
Best guess	These are paragraphs and not headers
Editor supplied	Confirm, edit or change the best guess add new
knowledge	information where necessary
Example case	h3 tag is used for formatting - content is too
(HTML	long - more than 100 characters(may be changed)
description)
Example use,	replace the header tag with p and class name we
WCAG	need to create CSS classes h1-h6 for each heading level
description

[0291]

TABLE A-41
Field	Description
Dependency	3.5.6
Transformation	H-3
Number
Knowledge gap	Structural role of the paragraphs
Logic	header tag used for formatting
Used for	replace header tag used for formatting with p tag using
	bold (options may be p with stylesheet or blockquote
	tag).
About	other catches header tag used for formatting
Best guess	These are paragraphs and not headers
Editor supplied	Confirm, edit or change the best guess add new
knowledge	information where necessary
Example case	Case 1: h4 tag is used for formatting - content is too
(HTML	long - more than 100 characters(may be changed)
description)
Example use,	replace the header tag with p and class name we need
WCAG	to create css classes h1-h6 for each heading level
description

[0292]

TABLE A-42
Field	Description
Dependency	3.5.7
Transformation	H-3
Number
Knowledge gap	Structural role of the paragraphs
Logic	header tag used for formatting
Used for	Replace header tag used for formatting with p tag using
	bold (options may be p with stylesheet or blockquote
	tag).
About	other catches header tag used for formatting
Best guess	These are paragraphs and not headers
Editor supplied	Confirm, edit or change the best guess add new
knowledge	information where necessary
Example case	Case 1: h5 tag is used for formatting - content is too
(HTML	long - more than 100 characters(may be changed)
description)
Example use,	replace the header tag with p and class name we need
WCAG	to create css classes h1-h6 for each heading level
description

[0293]

TABLE A-43
Field	Description
Dependency	3.5.8
Transformation	H-3
Number
Knowledge gap	Structural role of the paragraphs
Logic	header tag used for formatting
Used for	Replace header tag used for formatting with p tag using
	bold (options may be p with stylesheet or blockquote
	tag).
About	other catches header tag used for formatting
Best guess	These are paragraphs and not headers
Editor supplied	Confirm, edit or change the best guess add new
knowledge	information where necessary
Example case	Case 1: h5 tag is used for formatting - content is too
(HTML	long - more than 100 characters(may be changed)
description)
Example use,	replace the header tag with p and class name we need
WCAG	to create css classes h1-h6 for each heading level
description

[0294]

TABLE A-44
Field	Description
Dependency	5.0.1
Transformation	H-3
Number
Knowledge gap	What is the base role of the table
Logic	check conditions for data table
Used for	make a control rdf statement for each distinct table
About	Data tables need to be identified and known. Indicates for a data table include
	a table with borders, a table with cells that contain numbers, or short
	text without links etc . . .
Best guess	This is a data table
Editor supplied	Confirm, edit or change the best guess
knowledge
Example case	Case1: One case of data table is if table doesn't contain any other
(HTML	tables, has more then two column and rows, and content of each td is
description)	less than a threshold. td tag inside first tr tag of the data table
	Case2: Another case of data table is if table's border attribute is set
	td tag inside first tr tag of the data table.
	Case3: Another case of data table is if there is at least one td in the
	table with just a number in it td tag inside first tr tag of the data table.
	Case4: This case is when there is at least one th in the table. If so, we
	want to confirm that this is a data table.
Example use,	The action will create a special “control” action in rdf statement to
WCAG	reference another transformations that are dealing with the same
description	data table. And also will add a temporary dataT attribute, that will
	identify this table as data table

[0295]

TABLE A-45
Field	Description
Dependency	5.1.1
Transformation	H-3
Number
Knowledge gap	Table structure
Logic	td tag in the first tr of data table (must be header) The substitution of
	td cell to th cells is done only if there are no th cells in the table at
	all. Otherwise, if there is something in the table, it means that the
	author intentionally left some td's instead of th's.
Used for	identifying table headers
About	if we know a table is a dtata table then we should identify what is
	it's structure and what cells are headers. The first cell is typically a
	header. (must be header)But we are only confident of that if there
	are no cells marked as a header in the table at all. Otherwise, if there
	are some th (header) in the table, it means that the author intentionally left
	some td's (non headers) instead of th's.
Best guess	First columns act as headers
Editor supplied	Confirm, edit or change the best guess add new information where
knowledge	necessary
Example case	Case1: td tag inside first tr tag of the data table.The table is missing
(HTML	th tags One case of data table is if table doesn't contain any other
description)	tables, has more then two column and rows, and content of each td is
	less than a threshold.
Example use,	replace td with th. If there is a td tag in first tr in the table, we need
WCAG	to delete it and put a th tag instead with scope = “col” and content
description	remains as it was in td tag

[0296]

TABLE A-46
Field	Description
Dependency	5.1.2
Transformation	H-3
Number
Knowledge gap	Table structure
Logic	First td tag in not first tr (must be header) This transformation is
	dependent on the previous one, ie for sure there are th's in the table,
	either inserted by the author or the previous one. We need to add row
	headers in this transformation only if the column headers were added by the
	previous transformation and not by the author. We check it (for now)
	by looking at the first th in the first tr - if it contains scope attribute, we
	decide to add row headers. This algorithm may change in the future.
Used for	identifying table headers
About	first cell in each column (may be header) This transformation is
	dependent on the previous one, ie for sure there are th's in the table,
	either inserted by the author or the previous one. We need to add
	row headers in this transformation only if the column headers were
	added by the previous transformation and not by the author. We
	check it (for now) by looking at the first th in the first tr (cell) - if it
	contains scope attribute, we decide to add row headers. This
	algorithm may change in the future.
Best guess	First row act as headers (user need to confirm and add any rows that
	act as headers)
Editor supplied	Confirm, edit or change the best guess add new information where
knowledge	necessary
Example case	Case1: first td tag inside non-first tr tag. Table doesn't have any th
(HTML	tags.
description)
Example use,	Replace td with th. If there is a first td tag in non-first tr in the table,
WCAG	we need to delete it and put a th tag instead with scope = “row” and
description	content remains as it was in td tag

[0297]

TABLE A-47
Field	Description
Dependency	5.2.1
Transformation	H-3
Number
Knowledge gap	Association of data table cells with headers
Logic	If none of column table cells have scope or headers attributes, need to
	assign scope attributes for header cells.
Used for	Associate data table cells with headers.
About	We need to define what cells fall under what header cells if none of column
	table cells have scope or headers attributes, need to assign scope attributes
	for header cells. Our first guess will be the headers under which each cell falls
	(vertical and horizontal) is its ‘headers’.
Best guess	Table cells are logically under the any column headers that they fall
	under visually
Editor supplied	Confirm, edit or change the best guess add new information where
knowledge	necessary
Example case	Case1: for th tag with no scope attribute in the data table, add scope
(HTML	attribute to the th
description)
Example use,	add scope attribute with value col to the th
WCAG
description

[0298]

TABLE A-48
Field	Description
Dependency	5.5.1
Transformation	H-3
Number
Knowledge gap	Summary of a data tables
Logic	check the conditions for the data table,
	and if it doesn't contain a
	summary attribute, add one
Used for	assigning summary attribute for data tables
About	A complex data table, needs a summary attribute,
Best guess	Ask editor for the summary
Editor supplied	Table summary
knowledge
Example case	Case1: If there is a complex data table
(HTML	without a summary attribute,
description)	need to provide a summary
Example use,	add summary attribute with data table
WCAG
description

[0299]

TABLE A-49
Field	Description
Dependency	5.5.2
Transformation	H-3
Number
Knowledge gap	Table title and caption
Logic	for data table without caption child tag and without title attribute,
	but with a simple text right after it, need to provide either one
Used for	provide caption tag or title attribute for every data table
About	A data table should have a caption or title associated with it, Often a
	caption need has been provide in the page, but not pragmatically marked
	as a caption
Best guess	Associate short simple text after a table as a caption
Editor supplied	Confirm, edit or change the best guess add new information where
knowledge	necessary
Example case	Case1: table tag with no caption child tag, and the first node after
(HTML	the table is simple text, that is probably instead of caption
description)
Example use,	add caption tag, put the simple text that is after the table as a content of
WCAG	caption. Also, add title attribute for a table. The user will choose either
description	to do caption or title

[0300]

TABLE A-50
Field	Description
Dependency	5.5.3
Transformation	H-3
Number
Knowledge gap	Table title and caption
Logic	for data table without caption child tag and without title attribute, but
	with a simple text right before it, need to provide either one
Used for	provide caption tag or title attribute for every data table
About	A data table should have a caption or title associated with it, Often a
	caption need has been provide in the page, but not pragmatically marked
	as a caption
Best guess	Associate short simple text before a table as a caption
Editor supplied	Confirm, edit or change the best guess add new information where
knowledge	necessary
Example case	Case1: table tag with no caption child tag, and the first node before
(HTML	the table is simple text, that is probably instead of caption
description)
Example use,	add caption tag, put the simple text that is before the table as a content
WCAG	of caption. Also, add title attribute for a table. The user will choose
description	either to do caption or title

[0301]

TABLE A-51
Field	Description
Dependency	5.5.4
Transformation	H-3
Number
Knowledge gap	Table title and caption
Logic	for data table without caption child tag
	and without title attribute,
	need to provide either one
Used for	provide caption tag or title attribute for every data table
About	for data table without caption or title we need to get this
	information
Best guess	Ask editor to caption or title
Editor supplied	Confirm, edit or change the best guess add
knowledge	new information where necessary
Example case	Case1: table tag with no caption
(HTML	child tag, and the first node after the table
description)	is simple text, that is probably instead of caption
Example use,	add caption tag. Also, add title attribute for a table.
WCAG	The user will choose either to do caption or title
description

[0302]

TABLE A-52
Field	Description
Dependency	1.1.9
Transformation	H-4
Number
Knowledge gap	Role, equivalent or function of an image
Logic	image not inside a link with alt attribute
	with non meaningful content
Used for	provide a meaningful content for alternative text
About	Image not inside a link with a text equivalent with
	non meaningful content such as “click here”.
Best guess	3rd party OCR called by screen
Editor supplied	Confirm, edit or change the best guess
knowledge	add new information where necessary
Example case	Case1: image not inside a link with alt
(HTML	attribute with non-meaningful text
description)
Example use,	add notification of non-meaningful alt text
WCAG
description

[0303]

TABLE A-53
Field	Description
Dependency	1.1.1
Transformation	H-4
Number
Knowledge gap	Role, equivalent or function of an image
Logic	image inside a link with alt attribute with non
	meaningful content
Used for	provide a meaningful content for alternative text
About	image inside a link with a text equivalent with
	non meaningful content
Best guess	meaningful alternative text - that should
	probably be the role or title of the linked to page
Editor supplied	Confirm, edit or change the best guess
knowledge	add new information where necessary
Example case	Case1: image inside a link with alt
(HTML	attribute with non-meaningful text
description)
Example use,	add notification of non-meaningful alt text
WCAG
description

[0304]

TABLE A-54
Field	Description
Dependency	1.1.2
Transformation	H-4
Number
Knowledge gap	Role of each image map regain
Logic	area with alt attribute with non meaningful content
Used for	provide a meaningful content for alternative text
About	an Image map is a picture with different
	clickable regions. If there is a text equivalent,
	but it is not very meaningful then we may need
	more knowledge. Knowledge capture should
	probably be about were each regain links to
	(such as the title or role of a destination page
Best guess	Role of linked to page where each regain links to
Editor supplied	Confirm, edit or change the best guess
knowledge	add new information where necessary
Example case	Case1: area with alt attribute with non-meaningful text
(HTML
description)
Example use,	add notification of non-meaningful alt text
WCAG
description

[0305]

TABLE A-55
Field	Description
Dependency	1.1.11
Transformation	H-4
Number
Knowledge gap	Role of form control
Logic	input as an image with alt attribute with
	non meaningful content
Used for	provide a meaningful content for alternative text
About	A form control input button without text (such as
	an image) knowledge is missing as to what
	happens when the button is selected
Best guess	Typically the input type (such as submit) and
	form name and destination may provide help as to
	build this information
Editor supplied	Confirm, edit or change the best guess
knowledge	add new information where necessary
Example case	Case1: input (inside a form) is a graphical button with
(HTML	non-meaningful alt attribute
description)
Example use,	add notification of non-meaningful alt text
WCAG
description

[0306]

TABLE A-56
Field	Description
Dependency	1.5.2
Transformation	H-4
Number
Knowledge gap	Extra link may be needed
Logic	if area tag is used inside a client-side image map, and there doesn't
	exist any a tag with the same href as this area tag, need to add ‘a’ tag
	with redundant text link right after the map.
Used for	provide redundant text links for client-side image maps
About	Image maps can be hard to use, we need to know when there only
	way to get to a page is through an image map area.
Best guess	If a link to the same resource is not provided though a link tag then a redundant
	link may be needed
Editor supplied	Confirm, edit or change the best guess add new information where
knowledge	necessary
Example case	Case1: for each area tag, if there is no redundant text link for it on the page,
(HTML	provide one.
description)
Example use,	provide redundant text links for client-side image maps
WCAG
description