Hi Tom,
(I changed the subject line as it's just picking up one point from the main
thread)
>Regardless of the answer, I find myself puzzled by the
>implications of one assertion in the RDF schema:
>
> dcterms:alternative isDefinedBy http://purl.org/dc/terms/
>
>According to http://www.w3.org/2000/01/rdf-schema,
>"isDefinedBy" means "Indicates a resource containing and
>defining the subject resource". I am confused because
>I think of "alternative" as being uniquely _identified_
>by the string http://purl.org/dc/terms/alternative but
>actually _defined_ and documented in the Web resource
>http://dublincore.org/usage/terms/dc/ (to be precise, at
>http://dublincore.org/usage/terms/dc/#alternative-002).
I must admit I find rdfs:isDefinedBy somewhat slippery, but there's a
useful thread over in www-rdf-interest (mostly about the formation of
namespace names/URIs) but see especially:
[Aaron Swartz - on whether it is possible to derive the namespace
assocaited with a term from its URI]
http://lists.w3.org/Archives/Public/www-rdf-interest/2001Jan/0175.html
[Dave Beckett]
http://lists.w3.org/Archives/Public/www-rdf-interest/2001Jan/0185.html
[Aaron Swartz]
http://lists.w3.org/Archives/Public/www-rdf-interest/2001Jan/0204.html
[Dave Beckett]
http://lists.w3.org/Archives/Public/www-rdf-interest/2001Jan/0208.html
I don't think rdfs:isDefinedBy is _limited_ to establishing a relationship
between a class/property and the namespace, but it certainly seems useful
to use it in that way. (As I think Dave says in one of those messages that
URI might or might not resolve to a schema.)
If that relationship between resource and namespace is not expressed
explicitly, it is not visible in the RDF model. I think it is certainly
useful to have that relationship visible as a means of grouping terms.
(FWIW, I think it would be extremely dangerous to try to do this on the
basis of substrings of URIs. If it begins with "http://purl.org/dc/terms/"
then....)
So it seems a Good Thing to use rdfs:isDefinedBy in this way. I think I'd
suggest the pointer to a prose specification should be an rdfs:seeAlso
relationship.
But I must admit I would appreciate some clearer guidance on use of
rdfs:isDefinedBy.
Pete
|