W3C home > Mailing lists > Public > wai-xtech@w3.org > April 2005

two levels of processors

From: Al Gilman <Alfred.S.Gilman@IEEE.org>
Date: Fri, 15 Apr 2005 12:57:42 -0400
Message-Id: <p06110405be859babf575@[10.0.1.2]>
To: wai-xtech@w3.org


I want to expand a little on the 'partial understanding' theme as
regards what I think is the working assumption in PF.

<quote cite=
"http://www.w3.org/mid/p06110403be858963ac68@%5B10.0.1.2%5D">

.. two levels of understanding:
- standard terms in standard syntax
- KR in more general RDF (including the definitions of the standard terms)

</quote>

In other words, if there is a <link rel="x2:secondary" href="#foo"/>

[...]

<div class="portlet weather" id="foo"> ... </div>

Some processors will just use the 'secondary' concept because they
know the WCAG guidelines for accessible page structure say

IF
some element matches the semantics of the 'secondary' value for the
'role' attribute as defined in XHTML 2.0
AND
you are writing in HTML 4.01 or XHTML 1.0,
THEN
use a link with this 'rel' value and href pointing at the element
that fits the 'secondary' semantics.

These processors can recognize this case by string-matching
x2:secondary (or x2Role:secondary, or whatever we finally resolve to
be the recommended symbol) without parsing it into a prefix:suffix
pattern for namespacing and domain-blending.

Other processors will use the RDF ontology we publish for the
'secondary' concept together with metadata that the portal provides
keyed to the 'portlet' and that the newsML vocabulary binds to a
'weather' concept as composed knowledge to guide their [show, hide,
minimize, style, present] processing of this element.

These processors are set up to do more intensive processing on
knowledge, or the Semantic Web, if you will, and look more closely at
the text that they process.

Gottfried and Lisa took an action at the Tech Plenary F2F to sketch
out how this subclassing works; a) what the derivation model is in
terms of coarse-grain, more-standard object classes and fine-grain,
less-standard derived objects, and b) how we express this in syntax
which is as little disruptive as we can.

We are stuck with single-heredity (one role per entity) for the
coarse-grain model for compatibility with the established base in
assistive APIs.

I for one believe that a) we need tuple-space logic (multiple
heredity) for the fine-grain space of classes, and b) the 'class'
attribute is an appropriate place to use existing HTML 4.01 syntax to
refine the classes beyond what is available in the standard 'role'
values.

At a more abstract level, we want as much of the semantics
in our HTML techniques to be expressed in markup selectable
by XPATH [1.0?] expressions over XHTML 1.0 content as
possible.

Noah Mendelsson said "partial understanding" was his one-word answer
to how to resolve the challenges facing W3C today. I hope we can
interest him in our application, Because once again, assistive
technology is the fringe that is innovating where angels fear to
tread.

Al
Received on Friday, 15 April 2005 16:57:46 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 13:15:39 GMT