Hi Matthias,
But this is just philosophical.... In reality, in my daily working
life.... I refers to objects that have properties. I don't care if
one name it an infroamtion resource, a web document, a Thing, etc.
What I check is: is this URI in one of my triple store? No? Then can
I resolve this URI on the web? yes? Is there RDF? Yes? Is this RDF
describing this URI? yes? then lets do something with it!
In scenarios where you have the time to look at each resource and make
manual/mental disambiguations, this might work. This strategy does not
work very well for large-scale information integration, though. When
we want to be able to write queries, mappings and make inferences
spanning over large heterogeneous information resources, we cannot
afford to spend too much time for such disambiguations on a
case-by-case basis.
No, here I am referring to: is this URI representing something, or *is*
the thing. Anyway, as I said, I don't think we *ever* refers to the
actual "thing".
What I say is: the Web is just a mean to get the RDF description of
resource identified by an URI. But it is not because the URI is not
resolvable on the Web that no information exists for that URI.
My comment has nothing to do with the effectiveness of integrating large
scale data sources. But only about the meaning we give to an URI we use
to refers to a rdf resource :) (and its relation to the Web).
Take care,
Fred
