W3C home > Mailing lists > Public > w3c-wai-gl@w3.org > July to September 2001

Re: Data models?

From: Al Gilman <asgilman@iamdigex.net>
Date: Fri, 24 Aug 2001 16:47:56 -0400
Message-Id: <200108242028.QAA7046427@smtp2.mail.iamworld.net>
To: w3c-wai-gl@w3.org
At 04:03 PM 2001-08-24 , you wrote:
>>Is it possible to encode content structure into a script? How would this be
>addressed? I like data model because I have a pretty good feel for what we
>mean (though I obviously can't articulate it). But I doubt that others are
>going to have the same feel for it. We need something a bit clearer.
>
>Thoughts anyone?

The PF working group has been giving this a lot of thought and is on the verge
of asking you all to look at what we have come up with.

But don't get your hopes up too far.  We don't get much more precise than
saying 'data model.'

For example, we went around long and hard on little-s schemas vs. captal-S
Schema, the latter being XML Schema from W3C.  It turns out that different
schema languages can be better for different things, and the simplicity of
schemata that one can write in TREX, for example, may be critical for assitive
applications.  So we backed off from saying "use W3C brand schemas" at least
for now.  Not that many people shouldn't use the W3C language, but there is
insufficient cause to say conclusively that Schema should be used
_exclusively_
to schematize XML applications.  Enough technical drivel.

Please review and discuss with PF (there will be a public discussion
mechanism)
the XML Guidelines that are about to be floated for wider comment before going
down this path.

A data model as used here, is a more formal meaning than "markup or
equivalent"
even though markup _is_ used to bind content into a data model.  But there is
some resizual fuzziness [not a 100% crisp and unique formal sense to 'data
model' as used here] because different ways of composing or recording such a
data model have [formally recognizable] different characteristics and
capabilities.

But 'data model' as used here means some way of defining structural and
categorical generics, categories and structures that one can use to classify
and organize their web content.

To be effective, these purported generic qualities do have to be bound somehow
to the specifics of your content, and in ways that user agents [both
mainstream
and as augmented with assistive technology] can key off to implement graceful
transformation.

But that's what the XMLGL is all about.  Read it and let's think about
where to
go next.  [coming soon to a website near you...]

Al

what you document in a schema.

At 04:03 PM 2001-08-24 , you wrote:
>In checkpoint 1.3 we refer to data models. What exactly do we mean by this?
>
>My guess is we mean "any method by which structure is defined other than
>markup." So "data model" is a catch-all, or, as I mentioned in the telecon,
>an "equivalent."
>
>We could use "equivalent" instead:
>
>1. the hierarchical structure of the content is unambiguously represented in
>the markup or equivalent.
>
>But that's a bit too general for my taste.
>
>So do we have any examples of instances where structure is encoded in
>something other than markup?
>
>One obvious one (to me) is when content is stored in a database. Some user
>agents might be able to pull this content out without any associated markup.
>In this instance, the data model of the database would provide the structure
>for the data. But this is a long way off from what we think of when we talk
>about structure in a Web page.
>
>What about content in XML? If I am creating not only the content, but an
>XML-based grammar for markup, then it is important that my XML-based grammar
>reflect the structure inherent in the content. So am I referring to a schema
>when I say "data model"?
>
>Schema can also be used WRT databases. If the above example were written:
>
>1. the hierarchical structure of the content is unambiguously represented in
>the markup or schema.
>
>Would this better reflect what we're trying to say?
>
>Is it possible to encode content structure into a script? How would this be
>addressed? I like data model because I have a pretty good feel for what we
>mean (though I obviously can't articulate it). But I doubt that others are
>going to have the same feel for it. We need something a bit clearer.
>
>Thoughts anyone?
>
>Chas. Munat
>  
Received on Friday, 24 August 2001 16:28:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:47:12 GMT