Re: @profile is wrong solution for indicating that RDFa is present

Hi Mark,

In your answer to Simone you also write:

[[[
However, to use this mechanism for RDFa is to put the cart before the
horse; if all we're interested in is the XSLT for GRDDLing, we're
effectively creating an 'empty' taxonomy to carry this information.
This won't be the first time I'm accused of pedantry, I'm sure  :)  but
we've just spent 4 years discussing whether this or that attribute
could legitimately have this or that meaning, and it doesn't feel
right to me to now say 'let's use @profile to indicate
transformations'.
]]]

And yes, in view of the HTML4.01 definition on @profile:

http://www.w3.org/TR/html401/struct/global.html#h-7.4.1

I agree this is a fair point!

[snip]
> I'm not for or against this, or any other particular solution, by the
> way. :) 

:-)

>         I'm simply saying that if the argument is that RDFa should be
> aligned with GRDDL (which is a little eleventh hour) then it should be
> done properly, rather than being used as an argument for overloading
> @profile. (Which as far as I can tell doesn't work out of the box with
> GRDDL anyway.)

??? AFAIK it does. Again as far as I know if you give a URI that is
GRDDL enabled to, say, to Virtuoso, then it would automatically process
the stuff extracting the RDF content. Ie, if we use @profile and the
document at that URI is 'prepared' then RDFa-d content are automatically
processed into triple stores. (Virtuoso being one of the main database
systems used behind, eg, dbpedia, so this can be potentially quite
important.) I wonder whether tabulator does not do the same thing
already (I am sure it would plan to do that.)

Having said that, if we have a clear follow-your-nose mechanism, than I
presume both Virtuoso and tabulator could include a direct access to one
of our online extraction engines relatively easily.

Very pragmatically, I think we do have two goals here:

- something for the follow-your-nose approach.
- piggybacking, as you say:-) on the very existence of GRDDL and its
deployment to ensure a relatively painless and quick deployment of RDF.
It is indeed 11th hours in some ways, but we should not ignore the
potentials of this.

What I have seen until now

- use DTD-s: does not seem to cut with various processors as a
follow-your-nose tool.

- use namespace for the follow-your-nose, and use (or let the user use)
the GRDDL @profile and explicit grddl transformations to access the XSLT
transforms: it does put more burden on the user, which is not
necessarily attractive.

However... I also wonder whether this is indeed 100% kosher. Does this
mean that we would have a separate RDFa namespace for @resource, @about,
etc? But is it then o.k. to use them in the XHTML code _without_ the
namespace prefix then? Until now the agreement was that these attributes
are _without_ a namespace, ie, the usage pattern we had until now is
fine. Wouldn't that radically change the issue if we are pedantic?

- use of namespace (and decorate the namespace document with GRDDL
attributes) to achieve both goals: beyond the problem with namespaces,
don't we also get to a different issue? Would the GRDDL work in (GRDDL)
practice if the RDFa document is served as text/html? 99% of the
documents are served as such and not as XML... In other words, wouldn't
that require to serve the RDFa document as XML? (If so, that would
become quite a hurdle for most of the users, even more than adding a few
more attributes...)

We could ask the GRDDL guys (eg, Harry) whether this is the practice or
not...

- use of @profile (and decorate the profile document with GRDDL
attributes) for both: would mean to swallow our uneasiness:-)

Any other possibility we can/should choose from?

Ivan

> 
> Regards,
> 
> Mark
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf