Re: on media types for OWL (5.13)

On Fri, 2002-10-25 at 18:43, Jonathan Borden wrote:
> Dan Connolly wrote:
> >
> > On Wed, 2002-10-23 at 21:36, Jonathan Borden wrote:
> > > Dan Connolly wrote:
> > [...]
> > > > So I propose that the reference document specify
> > > > application/rdf+xml as a suitable media
> > > > type for OWL KBs written in RDF/xml syntax.
> > > >
> > >
> > > I'd really like a better story about how such a media type says anything
> at
> > > all relevent to OWL.
> >
> > Hmm... yes, well, I think I'd like a better story too.
> > But I have spent quite a bit of time trying to tell
> > the story, and it doesn't get much better no matter
> > how hard I try.
> Actually, below we are getting somewhere.
> [snip]
> >
> > * to use app/xml is not to assert the content of
> > the document, at least not a strongly as app/rdf;
> > you might use it for test cases and such where
> > you don't really mean to assert the contents,
> > and you expect the namespace pointers, stylesheets,
> > or whatever you put in your document is enough
> > for your audience to figure out what they
> > need to know about it.
> >
> > * to use app/rdf is to buy into the RDF concepts
> > spec (cited from the app/rdf registration doc);
> > i.e. to explicitly license folks to add related documents
> > to the premises of arguments based on your document
> > (in particular, if this doc or some document
> > it rests on uses owl, the owl spec becomes
> > part of the premise of your argument).
> This is a good distinction. To use application/rdf+xml says that the
> enclosed statements are 'asserted' according to the the RDF model theory.

application/rdf+xml says that the RDF model theory applies,
plus any specifications of properties used in the document.

See, in particular,

"2.3 Meaning of RDF documents"

> This seems to argue for app/rdf _at the very least for Large OWL_. Ok you
> have me convinced that application/xml is not appropriate (application/xml
> indicates no particular semantics)


> > * to use app/owl is ... umm... hmm...
> >
> > I don't really see why anybody would choose app/owl;
> > maybe somebody who likes that idea could fill in
> > the blanks there?
> I would say that application/owl+xml would indicate that, for example, for
> Fast OWL, the OWL model theory were to be the defining model theory -- this
> could also be applied to Large OWL,

No, you get that just from app/rdf.

> the point being that the media type
> registration should indicate how the document is to be interpreted i.e.
> which semantics ought apply.

The semantics of all the properties used in an RDF document ought
to apply.

If you mix owl with, say, a mapping of SQL schemas into RDF,
the formal specification of the SQL properties applies
(i.e. can be used to justify conclusions) too.

> >Somebody would have to write
> > up an internet media type specification for app/owl
> > and get it reviewed in the relevant IETF fora
> > before it's all said and done. I'm not inclined
> > to do so.
> Hmmm... doesn't the TAG suggest that WGs draft media type specifications
> when appropriate?


> ... this is not to say that application/rdf+xml is not
> appropriate, but rather that if application/owl+xml is also appropriate or
> more appropriate in certain situations, we ought not be too worried about
> issuing an I-D/media type specification.

I'm not sure what you mean; there's no way to get around issuing
an internet draft if you want to deploy a new media type.

> I'd certainly be able to write one
> up if we decide to go that route.

Ah; good.

> How does the W3C suggest that WGs "get it
> reviewed in the relevent IETF fora" given that it is being suggested that
> WGs do this? Would an email from the WG to some email address be sufficient?

Yes, to the ietf-types mailing list. (not just one email; we'd have
to participate in the resulting discussion to the satisfaction
of the IESG.)

> > Current draft of the RDF app media type seems to be:
> >
> > hmm... doesn't cite the concepts spec yet...
> >
> I'd reference that, and reference the criteria for deciding when to use
> app/rdf vs. app/owl,

which criteria?

The only one I can think of is: if you want RDF/RDFS tools to grok,
you should probably stick with app/rdf.

> and reference the OWL MT. Not a long document I
> imagine.
> The main decision ought be: ought specs that define their own MT (model
> theory) register their own MT (media type).

Hmm... that doesn't seem to allow for mixing.
But I suppose I could live with specifying app/owl and seeing
how it gets used.

> Jonathan
Dan Connolly, W3C

Received on Tuesday, 29 October 2002 09:57:42 UTC