Binding AccessForAll Properties to schema.org Classes

One of the goals of Access For All (AfA) metadata is to improve the findability of resources that meet individual user needs. This page describes a proposal for extending schema.org classes with AfA-compatible properties. schema.org is a collaboration between major search engines to develop and collect shared microdata formats to improve search results.

The Access For All metadata standard was designed to be used in conjunction with existing metadata standards and specification. The intention is that existing metadata can be supplemented with additional properties containing the AfA information. schema.org defines classes for common resource types and is already being used to describe those resources. One of the recommended approaches to extending the specification is to add properties to existing classes.

The general way that schema.org microdata works is as follows:

the itemscope attribute is added to the container of the thing being described. (In the case of HTML5 videos, this would be the <video> element. In the case of images, a container must be placed around the <img> tag.)
the itemprop attribute is added to information within the container. Typically, you might wrap the text of a name, for example, in a <span> with itemprop="name". The AfA information is not meant to be displayed. In this case, the <meta> element is used with the combination of itemprop and content attributes.

What would this look like in an Authoring Tool?

The primary resource types currently of concern within an OER authoring tool include images, videos and the document as a whole.

Notes on AfA metadata

a value of false for an is- or has- property implies that it is known definitively that the resource "does-not-" or "has-not-." If it cannot be known definitively one way or the other, the property should be omitted. In the comments below, I'm considering things we can infer automatically, not things an author can specify through direct knowledge.

Sample Image Markup

<div itemscope>
    <img src="http://foo.org/img.png" alt="alt text here"/>
    <meta itemprop="has-alt-text" content="true"/>
    <meta itemprop="has-long-description" content="false"/>
</div>

Sample Video Markup

<video itemscope src="http://foo.org/img.png">
    <meta itemprop="has-caption" content="true"/>
    <meta itemprop="has-transcript" content="true"/>
</video>

Sample Document Markup

<body itemscope>
    <meta itemprop="is-mouse-accessible" content="true"/>
    <meta itemprop="is-keyboard-accessible" content="true"/>
    <meta itemprop="is-display-transformable" content="font-size font-face foreground-colour background-colour"/>
    <meta itemprop="has-hazard" content="false"/>

    ...
</body>

Full Property List

Properties on CreativeWork (a parent class of articles, images, videos, etc)

Schemas use inheritance (multiple, in fact). Properties defined on CreativeWork will be inherited by ImageObject, VideoObject, etc.

Property	Expected Type	Description	Comments on application to OER Authoring Tools
is-mouse-accessible	boolean	The resource can be controlled using the mouse alone, if necessary	Assume always true for OER-as-a-whole
is-keyboard-accessible	boolean	The resource can be controlled using the keyboard alone, if necessary	At OER level: always true; At per-video level: true if our player, unknown if Flash. Summary should combine all.
is-display-transformable	Text	vocabulary: font-size, font-face, foreground-colour, background-colour	At OER level: all present; At per-video level: all present for our player, none present if Flash/youtube. Summary should combine all.
has-eBook	boolean	an eBook alternative is available	Known false at OER level.
uses-colour-coding	boolean	The resources uses colour to convey information.	Unknown at all levels.
has-hazard	Text	vocabulary: flashing, sound, olfactory, motion simulation	Known to have NO hazards at OER level and at image-level; unknown at video level. Summary should combine all.
has-text-representation*	boolean	a textual alternative is available	Normally, this is in reference to a specified original access mode. For OER authoring tools, we should use the refinements applicable to more specific media types, such as "captions"
has-audio-representation	boolean	an auditory alternative is available	False at OER level, unknown at image and video level.
has-tactile-representation	boolean	a tactile alternative is available	False at all levels
has-visual-representation	boolean	a visual alternative is available
has-audio-description	boolean	an audio description of the visual content is available
has-caption	boolean	a text caption of the auditory content is available
has-e-book	boolean	an e-book alternative is available

Properties on ImageObject

Property	Expected Type	Description	Comment on application to OER authoring tools
*has-alt-text	boolean	this resource has alt text available
*has-long-description	boolean	this resource has a long description available	I'm not sure if we can know this one.

Properties on VideoObject

Property	Expected Type	Description	Comment on application to OER authoring tools
*has-transcript	boolean	this resource has a transcript available
has-sign-language	boolean	a sign-language translation of the auditory content is available	We can't know this one definitively.

Where Properties Are Recorded In Web Content

Note: If something is "unknown," the metadata property should be absent (not false, not empty).

Note: Even if properties are always unknown, that fact will be summarized and so OER should be scanned for the property anyway.

* = scanned and summarized in "General" section of display
** = not sure how to summarize this one

On the OER describing the text, not including images or videos

Property	Assessment
*is-mouse-accessible	always true
*is-keyboard-accessible	always true
*is-display-transformable	always all values
*has-eBook	always false
*uses-colour-coding	always unknown
*has-hazard	always present but empty

On Images

Property	Assessment
*uses-colour-coding	always unknown
has-alt-text	true or false
has-audio-representation	always false (technically, it's unknown, but false for this iteration)
has-long-description	always false (technically, it's unknown, but false for this iteration)

On Audio Files

Property	Assessment
has-transcript	always false(technically, it's unknown, but false for this iteration)
has-visual-representation	always false (technically, it's unknown, but false for this iteration)

On Videos

Property	Assessment
*is-mouse-accessible	always true
*is-keyboard-accessible	true for our video player, unknown for youtube/Flash
*is-display-transformable	all values for our video player, present but empty for youtube/Flash
**has-caption	true or false depending on UniSub query; note that caps in same language are "captions", those in different language are "subtitles" and count as "translations" in the summary
has-transcript	always false (technically, it's unknown, but false for this iteration)
*uses-colour-coding	always unknown
*has-hazard	always unknown
has-audio-description	always unknown
has-visual-representation	always unknown