> I indeed get the feeling that the distinction is very subtle, and
> the question is: does it matter in practice, or is this an argument
> of theory and principle?
Well the problem with subtle distinctions in any system is they can become
very large distinctions down the road.
The most important question for practical use is "will a query return the
right results". Unfortunately the answer is probably "no". If I query the
example RDF at section 2.2.5 of dcq-rdf looking for an x for which x
<dc:identifier> "http://dublincore.org" then I'm not going to find one.
Worse, because of the other mistake in that example where it uses
rdf:resource where rdf:value should be used then if I have merged in the
reasonable <rdf:resource> <rdfs:subPropertyOf> <dc:identifier> and produced
an RDFS closure of the graph (to make my query more likely to succeed) then
rather than get nothing I could get the completely wrong item returned,
namely the RDF list item it is attached to).
|