On Fri, 24 Sep 2004 10:25:05 +0100, Andy Powell <[log in to unmask]> wrote:
>Interesting to use this to browse DC-related ontologies... starting at the
>following URLs and working outwards (as it were!)
>
>http://www.w3.org/2004/ontaria/basic?focus=http%3A%2F%2Fpurl.org%2Fdc%2Felements%2F1.1%2F
>http://www.w3.org/2004/ontaria/basic?focus=http%3A%2F%2Fpurl.org%2Fdc%2Fterms%2F
>
>If nothing else, it highlights the somewhat odd dc:titles that we've
>assigned to our namespaces in the RDFS declarations! :-)
From the description provided alone, I've always found it difficult to
fathom whether
http://purl.org/dc/elements/1.1/
refers to
(i) the set of terms, the "DCMI Namespace", the "vocabulary" in the terms of
the RDF Primer, or
(ii) one representation of that vocabulary, the document in which the terms
of the vocabulary are described.
The absence of typing of the resource makes it difficult to decide and, as
you say, that title literal is extremely "slippery"!
The fact that
http://purl.org/dc/elements/1.1/
is the object of an rdfs:isDefinedBy triple doesn't help any because the
rdfs:range of rdfs:isDefinedBy is rdfs:Resource, and you can't conclude that
the target resource is a document.
The clause in the title literal "providing access to its content by means of
an RDF Schema" suggests (to me) that it is the document, and _not_ the
vocabulary, but it is still unclear. Perhaps more so, the use of dc:language
also suggests that it is the document: I'm happy with the notion that the
content of the document is in English, but not that the content of the DCMES
vocabulary is in English.
It is perfectly reasonable that DCMI may wish to describe both of these
things (the vocabulary and the document) - especially, I imagine, if, as
we've mentioned in the past, we want to provide descriptions in different
languages in different documents, in which case the _same_ vocabulary may be
described in _multiple_ documents, and each of those documents may have a
different language, a different agent as its dc:creator, a different
last-modified-date and so on.
And if we want to make distinct statements about the document and the
vocabulary, we need different URIrefs for those distinct resources.
The Nokia vocabulary server
http://swdev.nokia.com/uriqa/query.rsp
adopts this approach and for DCMES uses the URIref
http://purl.org/dc/elements/1.1/
to denote (I think) the document/representation
http://swdev.nokia.com/uriqa/?uri=http%3A%2F%2Fpurl.org%2Fdc%2Felements%2F1.1%2F&format=text%2Fhtml&naming=label&inference=exclude
And the second URIref
http://purl.org/dc/elements/1.1
to denote the vocabulary. (That resource is explicitly typed).
http://swdev.nokia.com/uriqa/?uri=http%3A%2F%2Fpurl.org%2Fdc%2Felements%2F1.1&format=text%2Fhtml&naming=label&inference=exclude
(Note - no trailing slash on that second URIref! - when I first looked at
this eighteen months ago, it took me ages to notice that, and I was left
scratching my head at how on earth the distinction between the two resources
was being made!)
We can quibble about the choice of URIrefs (e.g. we might use
http://dublincore.org/2003/03/24/dces
for the document), but the approach seems sound and goes a long way to
removing any ambiguity.
Pete
|