- From: Ginsberg, Allen <AGINSBERG@imc.mitre.org>
- Date: Mon, 12 Feb 2007 17:25:52 -0500
- To: <kifer@cs.sunysb.edu>
- Cc: <public-rif-wg@w3.org>
> You didn't divulged too many details, but if what you are proposing
makes
> any sense, then a large number of idiots has been working on the
problem
> for 40 years for nothing. This includes John McCarthy, Vladimir
Lifschitz,
> Mel Fitting, Ray Reiter, and I just scratched the top of the list.
> (Sorry for the sarcasm.)
All I am saying is that logic tells me what follows from what. If you
say that "p" follows, in some sense, from "p := not q" then you are
not using "not" in the same way as logic does.
I never meant to imply that the people working on these formalizations
of reasoning (as opposed to logic) are wasting their time - I did say I
was playing "devil's advocate". But I don't think those formalisms
offer the only approach to dealing with these issues. And it very well
could be that, in the end, i.e., at the application level, solutions
based on pragmatics are always required (or better).
> > In what sense what you have in mind is more classical than, say,
the
> > stable model semantics?
>
> As far as I understand it, stable model "semantics" is basically a
> procedural add-on to classical semantics involving an implementation
of
> the closed-world-assumption. It is, if you will, a way of
implementing
> the assumption that everything that you know nothing about is false.
> Classical semantics makes no such assumption.
> Your understanding doesn't come from reading papers on this subject
then.
> Or, if it does then you completely misunderstood these papers.
That is entirely possible. I am not an expert on that. But from what
I have read (including the Fitting survey you referenced) it does seem
to be a way of formalizing the closed-world-assumption, and that
assumption is not consistent with classical semantics, i.e., logic.
Allen
Received on Monday, 12 February 2007 22:26:03 UTC