Re: Atom and RDF

On Mon, 11 Oct 2004 20:39:11 -0700, Joshua Allen <joshuaa@microsoft.com> wrote:
> > happening (i.e. give a discription of what possible
> > representations could be obtained for a given resource, and
> > in what cases) RDF would be the natural source.
> 
> Yeah, I actually meant to refer to the "Accepts" header, which is one of
> many examples of useful metadata passed in HTTP headers.  HTTP has a
> metadata and extensibility story and doesn't use RDF -- and that is not
> worth losing sleep over.

Ok.

> > Taking RDF out of RSS just doesn't compare.
> 
> Danny was talking about Atom, and specifically about the use cases of
> using Atom as a transport for routing and publishing RDF.  Atom doesn't
> have any RDF to "take out".  And it's perfectly feasible to route and
> publish RDF using Atom or RSS 2.0 without having to modify them to be
> based on RDF.  We already have one syndication format based on RDF, and
> one based on XML.  What could we possibly gain by starting a holy war
> about whether the third should be switched from XML to RDF?  

Ok, could be wrong but I don't think Atom yet has a suitable means of
doing the equivalent of <enclosure>. So how would you feel about RSS
2.0 items as payload in Atom?

Maybe what you're suggesting (RDF as payload) is not such a bad idea,
but the main reason I'd like to see Atom inherit some of the
functionality RSS 1.0 has is for further decorating /descriptions/ of
the payload and the properties already associated with the payload.
You can already say the author of an entry is "..." but it would be
useful to add, they have a Personal Profile Document at "...". Or
perhaps a better example would be - the creator of the attached mp3
was "...".

(I would
> point out that a mapping between RDF and XML in the case of Atom is
> trivial, BTW)

Only if you limit Atom to the core terms. That's really the nub of the
issue - building Atom that way arbitrarily restricts the kind of
things it can say. Ok, as I think Graham said we might not want Atom
to be as promiscuous as RDF (I hope I remembered that correctly ;-).
But extensibility is in the charter.

It has been suggested that additional info could go into
foreign-namespaces elements. Unfortunately that suffers from the N^2
problem for interop, and doesn't in itself provide any systematic way
of associated more information with the entries in the feed. Mapping
to RDF would not be trivial.

As has been pointed out, each entry is given a URI, which means RDF
can talk about them. But then that raises the spectre of one (Atom)
feed for content, one (RSS/RDF) for metadata.Or perhaps following the
idea of RDF-as-payload we could have alternating content/metadata
(wierd, but there might be something in it...).

Cheers,
Danny. 



-- 

http://dannyayers.com

Received on Tuesday, 12 October 2004 08:34:22 UTC