Re: mappings between SWRL and Boley proposal

From: Hassan Aït-Kaci <hak@ilog.com>
Subject: Re: mappings between SWRL and Boley proposal
Date: Thu, 18 May 2006 12:14:50 -0700

> Peter F. Patel-Schneider wrote:
> 
> > I had in mind, in part, trying to explicate what I felt was missing from
> > the proposal and what was wrong with it.
> 
> Peter,
> 
> This is a commendable endeavor. Why, then, won't you try to be more
> constructive and less rhetorical? 

I have already said that the syntax isn't bad; that the idea of basing the
semantics on interpretations is unobjectionable; that the details of the
semantics are needed; and that the idea of requiring a substitution meaning
is likely to cause problems.  I seem to remember that I have expressed
distaste for multiple semantics as well.

> We all agree that the proposal is
> a sketch and must be amended. 

Amended how?  My worry is that we will go too far down a path starting at
the proposal and that we will have to painfully backtrack.

> You  seem to have some precise ideas
> about making it better. 

Not exactly.  I do have, however, some ideas on how to proceed, and they
don't have too much in common with the proposal.

> So why don't you contribute to it rather than
> take the stance of decrying it as a whole as a matter of principle?

But what is the "whole" to decry here?  As this is just "a sketch" it is
hard to determine what are the vital points of the proposal.

It it vital that there be a logic-inspired exchange syntax?  I'm not against
that.

It it vital that there be model-theoretic semantics?  I'm not against that.

It it vital that there be multiple semantics?  I am against that.

It it vital that there be a subsitution meaning at the heart of
conditionals?  I am against that.

> If you think that something in it is wrong, missing, or needs to be
> made clearer and more precise, it would benefit us all and this WG's
> objective to see you do so simply and constructively. If it is this
> proposal that you reject as a whole, then why don't you please make
> a counter proposal and enlighten the rest of us with what you see as
> a viable alternative.  

I had thought that the WG was not at the stage that technical proposals
were appropriate.  I had thought that the WG was still wrangling over use
cases, requirements, etc.

But, OK, here is a sketch of my proposal:

1/ The language of the RIF Phase 1 is function-free Horn clauses.
   The interchange syntax will be XML. 

2/ Reasonable XML Schema datatypes can be used, along with various built-in
   predicates over these datatypes.

3/ Either RDF or OWL can be used to provide information.  

4/ The semantics is a model-theoretic semantics similar to the semantics of
   RDF or OWL. 

5/ Compliance for formalism X will be measured as follows:
   a/ Partial mappings will be provided between the syntaxes of X and the
      RIF, including mappings between X's data language and RDF or OWL.
   b/ A subset of the RIF will be identified as being X-compliant.
   c/ For that subset the deductive behaviour of X must "mirror" reasoning
      in the RIF.

I guess that this is really two proposals - one based on RDF and one based
on OWL - but they do have a lot in common.  I, of course, would prefer the
OWL-based version.

> No offense, but the way you have acted on this
> issue thus far seems to me a tad childish and puzzles me. But perhaps
> it's just me feeling so ...

Well I didn't see any better way to again point out what I view as problems
with the proposal and the approach it hints at.

> -hak
> -- 
> Hassan Aït-Kaci
> ILOG, Inc. - Product Division R&D
> tel/fax: +1 (604) 930-5603 - email: hak @ ilog . com

Peter F. Patel-Schneider

Received on Thursday, 18 May 2006 21:01:37 UTC