- From: pat hayes <phayes@ai.uwf.edu>
- Date: Mon, 9 Jul 2001 17:03:53 -0700
- To: Ian Horrocks <horrocks@cs.man.ac.uk>
- Cc: www-rdf-logic@w3.org
>On July 9, Thomas B. Passin writes: > > [Ian Horrocks" <horrocks@cs.man.ac.uk> > > To: "David Martin] > > > > > What you really want to say is that that the class of > > > persons who buy diapers is exactly the same as the class of persons > > > who parented some other person. To do this you should use a > > > sameClassAs assertion along the lines of (using a pseudo-syntax): > > > > > > > I think it's a good idea to get questions like this set out as clearly and > > simply as possible before inventing new syntax or vocabulary for them. > >Sorry if the pseudo-syntax offended you - I wasn't really intending to >invent new syntax but just to abbreviate the DAML+OIL RDF syntax, which >is rather cumbersome for such exchanges as this (someone did promise >to come up with a "human readable" syntax, but we are still waiting). > > > If > > you were using predicate logic or KIF, I imagine you'd arrive at something > > like this: > > > > for all persons x it is not true that: > > ((x bought diapers) and (x is not a parent)) > >Using FOL is fine by me, although it has the disadvantage that the >translation into DAML+OIL isn't always obvious. However, I don't think >that what you have written does the trick - it certainly isn't >equivalent to the "pseudo" DAML+OIL I wrote. As I understood it, we were >trying to express that people buy diapers if and only if they are >parents. What you have written says that people buy diapers only if >they are parents. What you needed was to say: > >for all persons x it is not true that: > (((x bought diapers) and (x is not a parent)) > or ((x not bought diapers) and (x is a parent))) You could, but in KIF it would be a lot easier just to say (forall (?x)(iff (Bought-diapers ?x)(IsParent ?x))) But the key point is the one Ian brought up earlier in this thread: these sentences (in KIF or CGs or DAML+OIL) all say that the two predicates (or classes) have exactly the same instances. If true, they are true in *all* interpretations, not just about the particular data that has been recorded so far. If sentences like this are asserted and then someone finds a non-parent buying a diaper, they have generated a logical contradiction. I don't think that is what Martin really wanted to do. For the record, this issue - of wanting to distinguish between logical expressions of meaning, on the one hand, and mere contingent facts, on the other - recurs again and again in ontology work, and has been the subject of considerable discussion. Nicola Guarino and others have argued that a fully expressive ontology language needs modal notions such as necessity to distinguish 'rigid' properties from mere accidental properties, for example. I mention this only to suggest that DAML+OIL isn't likely to be able to keep everyone happy over this. 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
Received on Monday, 9 July 2001 20:03:58 UTC