Yes Erik - you are quite correct. In my argument I deliberately started
by using informal terminology in building the case that the values
of this subset of the DCMES elements should strictly be *identifiers*.
There is a real problem in that the canonical definitions in DC1.1
do not make this clear. If it were possible to revise these (probably
not) then the CCP elements definitions should also say "a reference to"
or better still "an identifier of" ... The User Guide and our other
documentation should certainly make this point very very explicitly.
"Jul,Erik" wrote:
>
> Simon:
>
> Thank you for your elucidating note. I have a question about one of your
> assertions:
>
> > Firstly, we need to understand that there is a subset of the
> > DCMES whose
> > values are themselves "resources": these are CCP, Relation
> > and Source,
> > and (since "place" is in the DCT1 list) possibly Coverage.
> > For each of
> > these properties, the value is another resource that may have its own
> > description.
>
> I think this is exactly where we start getting into trouble.
>
> According to the definition of Relation and Source, for example, the only
> allowable value for these elements is "a reference..."
>
> Now, I understand your assertion, namely, that "For each of
> these properties, the value is another resource that may have its own
> description," in RDF terms, but the DC element definition does not seem to
> allow for your RDF interpretation, that is, the actual resource may *not* be
> the value of the Relation element, for example, only a *reference* to a
> resource may be the value.
>
> Read the definition:
>
> Identifier: Relation
> Definition: A reference to a related resource.
>
> It does not say "or another resource that may have its own description."
>
> In the troublesome CCP elements, we made the mistake of defining the
> elements as "An entity ...," as though the value of the DC Element would or
> could somehow actually *be* the entity itself. Of course it cannot. At
> best, we can have a *reference* to the entity. In fact, that's all we can
> ever have.
>
> With that understanding, we are free to discover, list, create, or restrict
> the *types* of references that we wish to allow such as name, URL, ID
> number, or other.
>
> --Erik
--
Best Simon
|