RE: rdf inclusion

> From: Graham Klyne [mailto:GK@ninebynine.org] 
> Pragmatically, it seems that any number of programmers have 
> been able to 
> build things using RDF in ways that respect the simple 
> assertional nature 
> of RDF for the most part, but also add features as needed for 
> the purpose at hand.

Correct.  And it is *always* possible to build something that appears to be
appropriate for 'the purpose at hand', and it will *always* be easier than
building something general-purpose because you don't have to worry about the
special cases you are wiring in.  Often, you happen to know that some nasty
cases won't come up in the data with which you're dealing, for example.

> Rigorous logic seems to be saying that there is 
> something wrong 
> about this, but these implemented systems work (pretty much) as their 
> programmer-designers intended.  Who's right here?

Both groups.  If you want to hack around with RDF as a dead simple data
representation that doesn't mean anything, show other people those hacks,
and get general disagreement as to whose hacked-together program and whose
operational definitions to use, that's done --- and has been re-done many
times.  If you want agreement as to how to interpret some of this RDF data
in a more general way, that's harder and there are problems trying to use
(say) cwm as the basis for it.

		- Peter

Received on Friday, 24 May 2002 09:40:02 UTC