Re: What do the ontologists want

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Sent: Thursday, May 17, 2001 12:17 PM
Subject: Re: What do the ontologists want

> I find the amount of RDF that DAML+OIL uses to be quite instructive
> in this regard.  DAML+OIL does not use reification, nor does it use
> containers.

But I think it would do if it wanted to express some of the prose
semantics for terms in RDF. And then you get the problems, such as the
fact that "daml:DatatypeProperty" violates the "typed nodes can't
inherit range and domain semantics" rule and so forth... (as if anyone
cares). Incidentally though, I don't think that there are going to be
processors that try to bootstrap DAML in from nowhere - there's always
going to be some level of primitives. In fact, to infer some of the
DAML rules, you'd need a language that has some terms that are more
powerful than DAML itself, so that'd be pointless.

> I think that there are some of us who worry mostly about the
> theoretical aspects of RDF, and some of us who worry mostly
> (solely) about the practical aspects of RDF.  Unfortunately for
> the latter group, it is impossible to have a web representation
> language without have the former taken care of, all protestations
> of some in the latter group notwithstanding.

Well, I didn't mean to imply that the former wasn't important - just
that sometimes there is an annoying level of pedantry that isn't
needed for implementation. I think that this particular thread has
been quite enlightening though.

RDF is far from complete, but once we have a stable model and a
cohesive syntax, it shouldn't to build applications on top of this and
then leave the general purpose machines to sort it out. CWM has been
quite an inspiration to me.

Whilst I'm on the thread, I agree with Jim Hendlers comments that
getting rid of triples as is would be a bust. I also think that to get
rid of reification would be a bust, noting some of Bill deHora's
comments. In fact, DAML can be used to model systems that use
reification quite neatly, so why not? The annotation people will
probably find a use for it, and there are loads of applications for
the old ":p :says { :x :y :z }" thing.

Maybe the KR folk are just grumpy because they didn't think of
grounding their terms in a global information space... [sound of me
ducking].

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .

Received on Thursday, 17 May 2001 10:31:57 UTC