>
> It's the only view that I've been able to live with,
> though interestingly (and worrysome) is that RDDL is
> not compatible with that view IMO since a RDDL document
> does not actually enumerate the members of a namespace
> but rather describes *other* resources than the namespace
> which have some relation to the terms in the namespace.
>
> So a RDDL document is not, per se, a *representation* of
> a namespace, but metadata which simply references the
> namespace or individual terms in that namespace. Sigh...
You're right. RDDL is some metadata about related
resources.
The issue - if there is any - is just shifted to
the term "representation", which is not well defined
in this context.
>
> > Also...
> >
> > > A namespace is nothing but a set of names. A namespace includes
> > > *nothing* about how those names are used. A namespace imparts
> > > no semantics whatsoever to the names residing in that namespace.
> >
> > OK, but (I think) this paragraph is referring to "namespace" in the
> > sense of "XML Namespace".
>
> Correct.
>
> > I think my question yesterday was whether "DCMI namespace" as
> > defined in
> > [1] was (perhaps unintentionally!) in fact being used to mean
> > something
> > slightly different from an "XML namespace" i.e.
> >
> > - a collection of _terms_ (rather than just names), where a
> > term is a DC
> > element (not XML element), element refinement or encoding scheme (i.e.
> > not just a name-without-semantics but a concept-with-semantics )
> >
> > On this basis, I think the "semantic surplus" in the RDF/XML
> > representation is much less of a problem if that RDF/XML document is
> > regarded as a representation of a "DCMI namespace".
>
> True, though then you have ambiguity if the same URI denotes both
> an XML namespace and a DCMI namespace and hence this would not
> be compatible with the view above.
>
> It would only be compatible if we did not consider namespace URIs
> to denote the namespace itself, but only as punctuation, and that
> those URIs may denote anything whatsoever and the thing denoted
> need not have any relation to the actual namespace. In that case,
> the URI would denote a DCMI namespace (termset) and the RDF/XML
> representation would be fairly accurate, and that DCMI namespace
> URI would simply be used as punctuation as an XML namespace to
> differentiate/uniquify the terms in the DCMI namespace.
>
> Sheesh, what a mess...
>
> Patrick
>
> > Pete
> >
> > [1] http://dublincore.org/documents/2001/10/26/dcmi-namespace/
> >
> >
>
>
|