Re: What are Semantics? (Was: Serving generic XML)

The "prescriptive" approach of saying that everyone should use XHTML is one
approach.  However many people don't like being prescribed to!  And one size
doesn't fit all.

Another approach, perhaps more "enabling", would be to say that we recognise
that XML + CSS style only gives a UA _part_ of what it needs (it is not
sufficient for
accessibility, for instance), and that authors need some (machine-readable)
way of publishing what their XML tags mean.

UAs would be able to retrieve/cache this "XML Tag Description file" in the
background and would be able to determine definitively that in a specific
document "<headline>" is "a header" (provided that "a header" is one of an
agreed set of "meanings" managed by a central authority or organisation such
as the w3c).  The UA would then be able to provide the appropriate
accessible rendering.  (and clearly there are other applications too.)

All we would need is an agreed (presumably XML) syntax for this "Tag
Description File" and, crucially, an agreed set of standard "meanings".
This set of standard meanings could be far richer than the semantics that
XHTML supports.

Obviously, the more subtle the meaning, the less likely it is that it could
possibly be agreed upon.  But for meanings such as visual conventions like
"a heading", or non-visual meanings such as "a patient", "a dentist", "a
credit card number", "a longitude and latitude", "a temperature" or "a
flight number" it is pretty clear-cut and unequivocal what they mean.
Indeed we could have hierarchies of such meanings, or rather "ontologies".

But hang on, such a "Tag Description File" already exists!!!!  Isn't this
the kind of thing that RDF can/could do.

So, after that lengthy preamble, my point is that instead of going down the
XHTML route (which doesn't lead anywhere) why don't we "cut to the chase"
and go down the XML + CSS + RDF route.

RDF (or something similar) could be used by UAs to determine the "missing
knowledge" required to produce accessible renderings and in a whole host of
other applications, because it would indicate to UAs what the tags "mean".

Isn't that the whole point?

nik


----- Original Message -----
From: "Tantek Çelik" <tantek@cs.stanford.edu>
To: "Elliotte Rusty Harold" <elharo@metalab.unc.edu>; <www-tag@w3.org>;
<www-style@w3.org>
Sent: 19 August 2002 22:46
Subject: Re: What are Semantics? (Was: Serving generic XML)


>
> On 8/19/02 1:52 PM, "Elliotte Rusty Harold" <elharo@metalab.unc.edu>
wrote:
>
> >
> > At 1:29 PM -0700 8/19/02, Kynn Bartlett wrote:
> >
> >
> >> If you make an arbitrary XML document, you cannot expect a browser to
> >> determine that <headline> is supposed to be "a header" -- even if you
> >> have visual presentation added on (via styles) which show that.
> >
> > Ah, but I can expect exactly that. And if a browser fails to do so
> > then I say this is a flawed browser,
>
> LOL!
>
> You actually expect a UA to parse the English tag name "headline" and then
> conclude it is a header, and then make similar conclusions for all other
> valid XML tag names?
>
> This is because unambiguously parsing English and assigning meaning to
> English words is a solved problem right?
>
> Please do some homework on the state of AI and Natural Language Processing
> before making such ridiculous assertions.
>
> And never mind the fact that 90%+ folks in the world don't speak English.
> Add "i18n" reading to your homework as well.
>
> > especially when it comes to
> > accessibility.
>
> Previously in this thread you have said several provably false things
> regarding accessibility.  If you wish to add value regarding
accessibility,
> please add the following reading materials to your homework:
>
>  http://www.w3.org/WAI/
>  http://www.w3.org/TR/WAI-WEBCONTENT
>  http://www.w3.org/TR/UAAG10
>  http://www.w3.org/TR/ATAG10
>
> and the documents linked from:
>  http://www.w3.org/WAI/Resources/
>
> > If humans can recognize certain visual layouts as
> > headers, then I think we should teach our computers to recognize them
> > too.
>
> This is because computer vision is a solved problem right?  Again, more AI
> reading would help here, as I don't think you understand where the state
of
> the art is, nor how far it has to go.
>
>
> Tantek
>

Received on Monday, 19 August 2002 20:51:25 UTC