Re: SWSL declarative semantics

From: Bijan Parsia <bparsia@isr.umd.edu>
Subject: Fwd: SWSL declarative semantics
Date: Thu, 20 Nov 2003 21:47:48 -0500

> Hi folks,
> 
> I got permission from Michael Kifer to forward this to this list. It 
> seems to fit in with some of our discussions. Hmm. at first glance, I 
> thought it might support a "comments count" view, or my own point about 
> the gensym fallacy fallacy, but i see that it might be more against 
> axiomatic approaches.
> 
> Cheers,
> Bijan Parsia.


> Begin forwarded message:
> 
> > From: kifer@cs.stonybrook.edu (Michael Kifer)
> > Date: Thu Nov 20, 2003  6:01:24 PM US/Eastern
> > To: swsl-committee@daml.org (Semantic Web Services Language Committee)
> > Subject: SWSL declarative semantics
> >
> > As promised, here is are a few thoughts on the nature of the
> > declarative semantics that we need for SWSL. This is a summary of a
> > discussion that we had with Karl at ODBASE-03 in Sicily 2wks ago.
> >
> > Currently, the General Requirements section states that SWSL should 
> > have
> >     "declarative semantics, in the typical sense used in knowledge
> >     representation where the meaning may be expressed in a
> >     logical framework that establishes overall principles of what
> >     conclusions are sancitoned from a set of premises".

Hmm.  This requirement is rather ambiguous, as it could apply to a
proof-theoretic semantics as well as a model-theoretic semantics.  
Futher, establishing ``overall principles of what conclusions are
sanctioned from a set of premises'' could be statisfied by exhibiting a
program that actually performs the inferences.

> > We felt that in the SW environment, it is inadequate to have semantics
> > that simply sanctions conclusions. 

Agreed.

> > The problem is that a
> > model-theoretic semantics by itself doesn't guarantee that all users
> > have shared understanding of the language constructs and thus use the
> > language correctly, especially if the logic language at hand is not
> > sufficiently high level and its semantics is given though a complex set
> > of axioms.  

This doesn't seem right.  The purpose of model-theoretics is precisely to
provide ``shared understanding of the language constructs'' and this is why
a model-theoretic semantics is not, and is not supposed to look like, a
``complex set of axioms''.  Perhaps Michael meant to complain about
proof-theoretic semantics, a complaint I whole-heartedly agree with.

> > So, we believe that there is a need for an informal conceptual model
> > (not unlike conceptual modeling in databases) that closely corresponds
> > to the human perception of objects, classification, processes,
> > etc. (e.g., UML-like). 

Well, sure, there should be such an informal model.  This is what a
model-theoretic semantics attempts to formalize.  (I would not, however,
use UML as an exemplar here, except as an exemplar of what can go wrong
when an informal conceptual model neither closely corresponds to human
intuitions nor is backed up with a comprehensible formal semantics.)

> > The language should then have constructs to
> > represent the concepts of that model directly and the formal
> > model-theoretic semantics of these concepts should be natural. By
> > "natural" we mean that a reasonable technically competent person should
> > agree that the formalization seems to adequately reflect the informal
> > semantics behind the conceptual model.

Agreed.  It is possible to create a model-theoretic semantics that diverges
from intuitions, which is not a good idea.

> > In other words, we need to make sure that there is a path from informal
> > human model of a particular task at hand down to the bowls of the
> > formalism that underlies the reasoning engine. Human knowledge
> > engineers are not going to verify their programs using formal
> > semantics, and in most cases they won't even fully understand it. By
> > providing a transition path from the informal to the formal we can gain
> > some confidence that the users' informal use of the language is
> > reasonably correct.

Agreed.  If the formal specifications don't match intuitions then they need
to be changed.

> > This approach applies to Semantic Web in general, not just SWS.

Agreed.  

> > The above should probably be an objective rather than a requirement.
> >
> >
> > 	--michael

Peter F. Patel-Schneider

Received on Friday, 21 November 2003 08:07:33 UTC