Re: RDF

Hey David,

On Mon, Jan 27, 2003 at 01:05:47AM -0500, David Booth wrote:
> Hi Mark!
> 
> I'm not entirely sure I've understood your comment (below), but if I did . . .
> 
> On Fri, 24 Jan 2003 01:09:17 -0500 Mark Baker <distobj@acm.org> wrote in 
> http://lists.w3.org/Archives/Public/www-tag/2003Jan/0336.html :
> >On Thu, Jan 23, 2003 at 11:32:26PM -0500, Tim Berners-Lee wrote:
> > > This (fragid hiding) is a form of end to end property.
> > > The significance of the the mime message is known only
> > > to sender/publisher and to receiver/browser.
> > > It is only in that context that the fragment ID makes sense.
> >
> >I understand what you're getting at, but I don't believe this is
> >particularly valuable.  You argue that "RDF uses this hook to introduce
> >identifiers for arbitrary concepts", yet Roy and I and others are
> >saying that you can already do that without this hook.
> 
> I believe some kind of syntactic convention is necessary, in order to 
> simultaneously (a) distinguish between the car and the picture of the car; 
> and (b) achieve the "View Source" effect.  I've explained my reasoning more 
> fully in
> http://www.w3.org/2002/11/dbooth-names/dbooth-names_clean.htm#EnablingViewSource

Yah, I've had a look through that.  It's pretty thorough, though I
believe the premise to be incorrect; there need be only one kind of
"thing".

Of the three kinds of things you've identified (I don't include "name"),
what's common about them all that we can break out into a new, more
general model?  Each has identity, and encapsulates some state.  Agreed?

So with this new model, those three previously disjoint things can be
unified.  This model, the REST model, says that;

- the concept of love is a "resource", an abstraction
- resolving an identifier for the concept of love will end up as a GET
request on some Web server/proxy (just to bridge with your "Web location").
- the resolution of that identifier will return documents that represent
the resource

The REST model is the most appealing of the models I've seen presented,
as it seems to be at least as powerful(*) as TimBL's, plus it's more
general, since http: URIs would be able to identify anything; love,
cars, people, etc..

(*) the implementation would need work to match TimBL's model, ala an
HTTP extension header called "ConceptualWork" whose value is a URI.
But the model seems adequate, if only because this extension is
possible.

MB
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
Web architecture consulting, technical reports, evaluation & analysis

Received on Monday, 27 January 2003 01:55:50 UTC