Matthew Cockerill wrote:
> That is one excellent way to dereference an ISBN.
>
> But it is onlyone of an indefinite number of choices. And so all
> developers who want to spot ISBNs in an incoming stream of metadata
> would have to continuously add to their list of 'possible URIs which
> might correspond to an ISBN'.
>
> But even doing so begs the question - what is this ISBN thing to
> which all these different URI forms correspond, and how can the
> developer tell that the IDs in these different metadata streams mean
> the same thing?
>
> A curated registry actually gives you a single thing to which all
> these URI forms can correspond. There is only one ISBN type within
> the URI registry.
>
> ISBN is a slightly unusual example, since it's one of the few
> identifiers to have been blessed with its own URN-space. which does
> that job already. But info makes it orders of magnitude easier to
> achieve the same for other identifiers.
IANARE (I Am Not A RDF Expert), but I thought the question "What does
URI X correspond to?" was answered by the "rdf:type" associations, i.e.
the "class" of URI is metadata about the URI. And those types are for
many common things quite well defined. In other words, the URI does not
contain type information since that is handled by the metadata. Or am I
missing something?
Also, IAAAE (I Am An AOP Expert), and in that world (which seems to
correspond quite well with the RDF mindset) it is utterly and completely
uninteresting what "class" or id a particular object has, in terms of
determining what one can do with it. Instead one focuses entirely on the
set of interfaces that are exposed by the aspects attached to it.
/Rickard
--
Rickard Öberg
rickard.oberg_at_senselogic.se
_at_work +46-(0)19-173036
Received on Tue Nov 15 2005 - 09:42:09 EST