- From: Bill Andersen <andersen@ontologyworks.com>
- Date: Wed, 11 Oct 2000 12:42:52 -0400
- To: Jim Hendler <jhendler@darpa.mil>
- CC: Dan Connolly <connolly@w3.org>, Grit Denker <denker@csl.sri.com>, www-rdf-logic@w3.org, seanl@cs.umd.edu
Jim Hendler wrote:
>
> While I'm not a details guy like Dan, I thought the definition of
> equivalency was a simple one (i.e. not actually equality) -- I
> thought we were striving for essentially "macro" like replacement --
> i.e. if (equiv XXX YYY) then it should mean replacing XX with YYY
> leaves meaning unchanged. This may seem silly, but consider it where
> one ontology has something called "automobile" and another has
> something called "car" -- I may want some agent to know it can map
> between these two so I would want to be able to assert that
> ontology1:autombile is equivalent to ontology2:car. Jeff Heflin and
> Sean Luke introduced this to the SHOE language, and it was quite
> useful (and they can correct me if I haven't explained it well) --
> Jeff's versioning work (see the SHOE web page) also uses this.
Hey all...
What you want is a version of Leibniz's Law applied to intensional
relations, in this case automobile and car, such that you can say
(-> (forall (P) (<-> (P o1:automobile) (P o2:car)))
(= o1:automobile o2:car))
Now, just what those properties P might be that apply to properties
like automobile and car I don't know. One might be that they have
the same extension. Also, your reasoner needs to take this equality
of properties into account. Note that all of this can still be done
in first order if your restrict yourself to *named* properties like
car and autombile (see Pat Hayes' rant on SO vs reification).
...bill
--
Bill Andersen
Chief Technology Officer - Ontology Works
1130 Annapolis Road, Suite 203, Odenton, MD 21113
andersen@ontologyworks.com / 410-674-7600
Received on Wednesday, 11 October 2000 12:42:02 UTC