- From: Graham Klyne <Graham.Klyne@MIMEsweeper.com>
- Date: Thu, 19 Sep 2002 11:39:03 +0100
- To: pat hayes <phayes@ai.uwf.edu>
- Cc: w3c-rdfcore-wg@w3.org
Pat, I don't have any problem with the proposed change, but I'm a little puzzled by your explanation. Looking at the words, I read "A rdfs:subClassOf B is true in I if ICEXT(I(A)) is a subset of ICEXT(I(B))" to mean the second part implies the first. Thus I don't recognize the possible concern that A's class extension being a subclass of B's class extension could allow A rdfs:subClassOf B to be false. #g -- At 04:01 PM 9/18/02 -0500, pat hayes wrote: >Guys, > >Recent attempts (in collaboration with Peter and Ian) to reconcile the >RDF(S) MT with the emerging OWL MT have suggested that it would be good to >make a technical alteration to the semantic conditions for RDFS. I think >that this change will make no appreciable difference to RDFS itself (ie it >will give the same RDFS entailments), but am soliciting comments on this >change from the WG before deciding to do it. > >We can illustrate the point by considering rdfs:subClassOf. Right now, we >say that > >A rdfs:subClassOf B > >is true in I if ICEXT(I(A)) is a subset of ICEXT(I(B)). Which seems kind >of obvious; but the idea is to change this to read " ... is true in I >IFF ... " , i.e. make the semantic conditions *necessary and sufficient* >for the truth of the triple. > >The reason this doesnt make any appreciable difference to RDFS is that >RDFS has no notion of negation, so it doesnt really matter if it treats >something as false when its really true, which would be possible in the >current MT: that is, you could (currently) have two classes and one class >extension be a subset of the other, and still count rdfs:SubClassOf as >being false in that interpretation. That wouldnt matter since that >interpretation wouldnt satisfy any antecedents that might trigger a wierd >conclusion, since it just fails to make something true. But when we go to >OWL, there is something like a negation (owl:complementOf), and so now >this possible mismatch between what is actually correct in the >interpretation and what triples the interpretation makes true suddenly >starts to matter, both ways round. > >Similar changes need to be made to the conditions for rdfs:subPropertyOf >and the domain and range conditions, for similar reasons. > >Unless anyone objects, I plan to make this change in the next (and >hopefully close-to-final) version. So object now or forever hold your >piece. (To emphasize, this change makes the MT *more* conventional rather >than less, ie this is the standard way to do it) > >Pat >-- >--------------------------------------------------------------------- >IHMC (850)434 8903 home >40 South Alcaniz St. (850)202 4416 office >Pensacola, FL 32501 (850)202 4440 fax >phayes@ai.uwf.edu http://www.coginst.uwf.edu/~phayes ------------------- Graham Klyne <GK@NineByNine.org>
Received on Thursday, 19 September 2002 06:23:32 UTC