RE: DAML+RDFS: potentials for simplifications?

> From: Seth Russell [mailto:seth@robustai.net]
> > From: "Peter Crowther" <peter.crowther@networkinference.com>
> > > Sure, it's feasible!   Follow the database records to
> > > where the rules are coded in a language that is
> > > understood by your coded in a language that is
> > > understood by your programs.  Here, this is one
> > > way you could do it:
> > > http://robustai.net/mentography/disjointWith.gif
> >
> > All this does is to shift the burden onto ever-more-complex 
> programs, of
> > whatever nature, that must then be embedded into all 
> > Semantic Web-aware clients.
> 
> There is no reason to embed them in *all* clients ... just 
> those that want
> to be semantically aware but not restrained by  the 
> limitations of DMAL.
> The path for finding the KIF rule is in no way complex; and 
> there are lots
> of programs that can draw inferences from these KIF rules, as 
> you well know.

Sure.  You're then saying that there will be a 'second class' of clients
that can't understand all the structure that is being presented to them.
We've already had this discussion several times at a lower level, as to
whether RDF- and RDFS-based clients could draw meaningful conclusions from
DAML; I think the conclusion is that it's meaningful as far as RDFS goes,
and that the result is not comparable to the DAML result.  Here, we would
have another layer, and more cases where the following dialogue happens:

Q: Is this Semantic Web aware?

A: It depends on what your idea of the Semantic Web is.

... not good for interoperability.

> >Unless, Seth, you have found a way around Godel's Incompleteness
> > Theorem [1]?
> 
> How is Godel's Incompletness Theorem involved here.  Could 
> you explain?

Sorry for not being clearer, but I was missing some information.  You hadn't
explained that this was limited to KIF.  I therefore assumed that arbitrary
extensions to the logic were desired.  If the complexity of the logic has
KIF as its upper bound, then that's fine... but why not simply specify KIF
rather than use some mechanism for extensions that some clients understand
and others don't?

		- Peter

Received on Thursday, 29 November 2001 13:52:08 UTC