Re: [OWL-S] Negative effects/delete lists

> > [Sheila McIlraith]
> >
> > I wonder if we can write a tool to ensure that negative effects coding
> > rules are not violated?
> 
> [Bijan Parsia]
> I believe so.
 ...
> Is it anything more than checking for each effect literal whether its 
> negation is entailed by the KB?
> 
There's a long history of work on this problem, which is really the
problem of figuring out how to find a coherent story about what
Strips-style action definitions actually _mean_.  It is easy to come
up with axiom sets for which deleting (or, for that matter, adding) an
assertion produces semantic fruit salad very quickly.  It's amazing
how little all this seems to matter in practice.  

I think the real question hiding in the bushes here is what kinds of
closed-world assumptions we can make in the semantic-web-service biz.
Reiter's book explains how to handle both closed and open initial
conditions within a clean logic framework, but in practice I tend to
assume that you need a lot of closedness assumptions for inference to
be tractable.

                                             -- Drew

-- 
                                   -- Drew McDermott
                                      Yale Computer Science Department

Received on Thursday, 1 April 2004 09:48:53 UTC