On Thu, Jun 15, 2006 at 01:13:54PM +0100, Pete Johnston wrote:
> > is very close to the URIs of the three namespace URIs
> > currently used for DCMI properties and classes, as described
> > in the DCMI Namespace Policy [2]:
> >
> > http://purl.org/dc/elements/1.1/
> > http://purl.org/dc/terms/
> > http://purl.org/dc/dcmitype/
>
> It is a different URI though ;-) i.e. there is no problem with
> applications distinguishing them.
...
> I don't think there is any technical reason for not using
> http://purl.org/dc/xml/ , but equally there's no requirement for the
> DC-XML format to use that URI and it would work just as well with a
> dublincore.org URI for the XML Namespace Name. So if "socially", that's
> an easier option for DCMI, I'm happy to change it.
Yes, that's my point. It looks too much like a DCMI namespace,
which could create the false impression...
> As an aside, one question I haven't quite got clear in my own mind is
> whether we should "version" the XML Namespace Name, and use something
> like
>
> http://dublincore.org/formats/dc-xml/2006/06/14/ (or whatever)
One simple solution would be to put up a document
(e.g., with embedded <a name="descriptionSet"> on the model of
http://dublincore.org/documents/dcmi-terms/#creator) and to version
that just like any other DCMI document. Implementors could decide
for themselves which URI to use, e.g.:
http://dublincore.org/formats/dc-xml/2006/06/14/
http://dublincore.org/formats/dc-xml/
> But short answer: yes, a dublincore.org URI instead of a PURL will be
> fine for the XML Namespace Name.
Okay.
> However, if DCMI ever did decide to define classes of DescriptionSet,
> Description etc, I'd suggest it would be at best very confusing to
> choose to use URIs like
>
> http://purl.org/dc/xml/descriptionSet
>
> based on the fact that DC-XML used the expanded name
> ("http://purl.org/dc/xml/", "descriptionSet"), because
>
> (a) the things you want to describe are the things defined by the DCAM,
> not the things used in the DC-XML format
> (b) you'd end up with the same QNames appearing in RDF/XML and in DC-XML
> for quite different purposes - and while that wouldn't be "wrong"
> because they would be interprted in the context of those two different
> XML languages, I think it would probably become confusing to a human
> reader
Thank you for articulating that - I got confused just thinking
about it...! :)
> I find it a bit harder to imagine a need to define classes for ....QName
> etc, as these are the names of things which are used only in the XML
> format, and I can't really see us needing to build RDF descriptions of
> an XML format or of XML instances -
Good.
> - other than as an academic exercise
> to fill our empty evenings when the World Cup is over ;-).
:)
Tom
--
Dr. Thomas Baker [log in to unmask]
Director, Specifications and Documentation
Dublin Core Metadata Initiative
|