Re: SEM: OWL Full semantics

>Pat, I better understand it now... Is
>http://lists.w3.org/Archives/Public/www-archive/2003Mar/att-0018/LBASE-new.html
>the latest state of the art?

Yes and no. There are actually two distinct styles of doing the 
translation, and this is one of them. The other translates eg 
rdf:subClassOf into a binary relation subClassOf, then has iff axioms 
to connect that to the 'logical meaning'. That second way gives a 
more accurate rendering of RDFS and OWL Full but includes the 'RDF 
clutter' that Ian and Peter dislike. The style in the above document 
is better suited to capturing the meaning of the OWL abstract syntax.

We are having some discussion about this right now on the SCL mailing 
lists (which one to use to map OWL into SCL?) I will argue for having 
both styles, and then to analyse the relationships between them in 
SCL where we have a proper logical proof theory and a full semantics.

Pat

>-- ,
>Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/
>
>
> 
>                                                                                                                     
>                     pat 
>hayes                                                                                         
>                     <phayes@ai.uwf.edu       To:     Jos 
>De_Roo/AMDUS/MOR/Agfa-NV/BE/BAYER@AGFA                       
>                     >                        cc: 
>www-webont-wg@w3.org                                             
>                     Sent by:                 Subject:     Re: SEM: 
>OWL Full semantics                                 
> 
>www-webont-wg-requ                                                                                
> 
>est@w3.org                                                                                        
> 
>                                                                                                                     
> 
>                                                                                                                     
>                     2003-05-12 
>08:37                                                                                  
> 
>PM                                                                                                
> 
>                                                                                                                     
> 
>                                                                                                                     
>
>
>
>
>
>>Pat, in your message
>>http://lists.w3.org/Archives/Public/public-webont-comments/2003May/0067.html
>
>>you wrote very sensible comments which I fully support.
>>If I may suggest, put on your OWL S&AS editor's hat and
>>put it those changes !-)
>
>AS&S is a multi-document, and each part of it has a different editor
>list. I only commented on the parts of the document where I am not
>listed as editor, and they were intended as comments to the editors
>of those sections.
>
>>   Further on, as I have suggested
>>many times, the OWL Full vocabulary extension of the
>>RDFS semantics [RDFS MT] could be made explicit in the
>>shape of corresponding entailment or inference rules
>>which are sanctioned by those OWL Full semantics.
>
>Yes, that could be done, if necessary as back-projections from the
>translation into Lbase, though it would be rather complicated to get
>a complete set of rules (and very complicated to prove it was
>complete), and they would not all be simple "closure" rules in the
>RDFS style.
>
>>What I captured from many people in the community
>>is that they are really interested in that and that
>>it would drastically increase interoperability between
>>OWL Full reasoners, verifiers and explainers. I also
>>believe that this is a realistic job; even achieving
>>a minimum set is much better than having nothing ;-)
>
>Interesting idea.  However, I think this would be a different
>document.  Obviously, if I had my druthers the entire AS&S document
>would have been written in a different style, with OWL presented as
>first an extension of RDFS (OWLFull, that is) then with a syntax
>restriction to get OWLDL, and with the abstract syntax and
>'conventional' model theory presented last, almost as an appendix.
>But this was clearly a minority view among the trio of editors; Peter
>and Ian had already done a great deal of work on the abstract syntax
>when we worked out the RDF-compatible semantics, and Ian was not
>fully convinced of the internal coherence of the RDF-style model
>theory; so to have tried to force my own ideas on the group would
>have led to endless in-fighting, and Peter has done a very nice job
>within the current style, so I don't think it is worth trying to
>rewrite the whole thing. I would just like to clarify the
>relationship of the various OWLs to RDF in the wording of the text
>here and there, is all.
>
>It might be worth trying to write an entirely different document in
>the style you suggest, by the way, a kind of integrated
>semantic/inference-rule account of all the W3C SW languages.  But if
>any of us suggest that this be done by Webont, Guus will turn purple.
>
>Pat
>
>
>--
>---------------------------------------------------------------------
>IHMC                                      (850)434 8903 or (650)494 3973
>home
>40 South Alcaniz St.                      (850)202 4416   office
>Pensacola                                           (850)202 4440   fax
>FL 32501                                       (850)291 0667   cell
>phayes@ai.uwf.edu                   http://www.coginst.uwf.edu/~phayes
>s.pam@ai.uwf.edu   for spam


-- 
---------------------------------------------------------------------
IHMC					(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.			(850)202 4416   office
Pensacola              			(850)202 4440   fax
FL 32501           				(850)291 0667    cell
phayes@ai.uwf.edu	          http://www.coginst.uwf.edu/~phayes
s.pam@ai.uwf.edu   for spam

Received on Monday, 12 May 2003 22:42:23 UTC