Roland,
> attached you find a rewording of the metadata, which i think you want.
Thanks, yes, I think that version makes a clearer distinction between
the metadata about the namespace/term set and the metadata about the
RDF/XML document.
I still think (see my exchange with Patrick) that given the current
definition of "DCMI Namespace", strictly speaking, it is a bit
problematic to identify an instance of that class using a URI which has
already been deployed as an XML namespace name (unless as Patrick
suggested, we treat XML namespaces always only as a syntactic device).
In the DCMI case there _is_ a one-to-one mapping between the names of
the terms in a "DCMI Namespace" and the names "in" the XML Namespace (if
you can really say that!), so maybe it becomes less significant... But
strictly speaking, the collection of DCMI terms is still a different
type of resource from the collection of names. And while I recognise
that in RDF a resource can have multiple rdf:types, I'm not sure that in
reality (!!), the same resource can be both a "DCMI Namespace" (as
currently defined) and an XML Namespace.
I really shouldn't speculate, but given the emphasis on XML Namespaces
in the opening section of
http://uk.dublincore.org/documents/2001/10/26/dcmi-namespace/
I'm not really sure that the authors really intended to create this
distinction. And maybe "DCMI Namespace" was meant to be a collection of
_names_ of DCMI terms, where that collection of names is identified by a
URI. i.e. a "DCMI Namespace" is just an XML Namespace named by DCMI, and
all DCMI Namespaces are XML Namespaces. And the problem of ambiguous
identification disappears. But I shouldn't speculate on that....! ;-)
> In practice the metadata should reside in a document, which
> is called by rdf:about via rdf:about="#" and by rdf:resource
> via rdf:resource="#" so that changes in the URL resolving the
> namespace URI do not result in a maintainance issue.
Agreed.
> The proposal does not change my opinion, that an XML namespace URI
> as such can be used to place term declarations.
> W3C itself is doing so for RDFS (mime-type text/plain)
> and in part for RDF (mime-type text/plain).
> The XML Schema namespace URI yields mime-type text/html
>
> From that i take, that namespace URIs can actually be used
> to place more or less useful documents.
Yes, fine with me, I think.
Pete
|