Re: thinking about the formal model for RDF

On Wed, 18 Jul 2001, Aaron Swartz wrote:

> On Thursday, July 12, 2001, at 07:28  AM, Stefan Kokkelink wrote:
>
> > There is no problem with anonymous resources here. It
> > is not said in sec 5 that a resource *must* be
> > named by a URI! Sec 5 just talks about the *set* of
> > resources. Sec 5 does not talk about a syntactic representation
> > of triples that, for example, a parser may *choose* to
> > represent the formal model.
>
> Well, yes, but earlier in the spec M&S states clearly that:
>
> 	"Resources are always named by URIs..."
>
> This is under the original definition of resource, towards the
> beginning of the spec.
>
> Something has to give,

This just means "URIs are RDF's preferred form of name", the only kind of
name that is "built in" in any sense to RDF. We shouldn't take it as
asserting that "We always know the (URI) name of any resource we're
concerned with", since that's patently false: Web apps always have to deal
with partial knowledge, we can't guarantee knowledge of all URIs.

For "Resources are always named by URIs", read "the RDF name for a
resource is its URI", not "we can't mention resources without providing
their URI".

Dan