Re: grounding terms in URI space

On Mar 11, 2005, at 8:33 AM, Miles, AJ (Alistair) wrote:
>> But I agree with Dan that having something useful there is a good
>> practice. And as he says, it is entirely orthogonal to the #
>> vs / debate.
>
> Yeah but, can I refer to some document that says:
>
> (1) If you use an HTTP URI of the form http://foo#bar to denote a 
> conceptual resource then the resource denoted by the URI http://foo 
> should accept HTTP GET requests and provide representations according 
> to content-types x and y (but not z) that convey information I.
>
> (2) If you use an HTTP URI of the form http://foo/bar to denote a 
> conceptual resource then the resource denoted by http://foo/bar should 
> redirect HTTP GET requests to another resource that provides 
> representations according to content-types x and y (but not z) that 
> convey information I.
>
> ... ?  If that's written somewhere (is it?), great, but if it's not 
> then the 'Quick Guide to Publishing a Thesaurus on the Semantic Web' 
> is defining (or at least extending) best practice recommendations, and 
> I think somebody else ought to be doing that first.

I'm not terribly concerned who in the Best Practices WG does it, as 
long as it gets
done.

And I don't think it's essential to go into the 2 cases above in the 
"Quick guide...". You already
say that you should publish an RDF representation of the thesaurus 
terms; just say that dereferencing
the terms should yield that RDF representation. If you want something 
to point
to, use webarch:

" A URI owner SHOULD provide representations of the resource it 
identifies"
http://www.w3.org/TR/webarch/#pr-describe-resource


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Friday, 11 March 2005 17:30:12 UTC