Re: equivalentTo

From: Ian Horrocks (horrocks@cs.man.ac.uk)
Date: 03/22/01


On March 21, pat hayes writes:
> >Hi:
> >
> >As I was going through the reference document last night, I again
> >encountered the issue of equivalentTo.
> >
> >equivalentTo has been used to make two things the same.  However, there are
> >several subrelationships of equivalentTo, including sameClassAs and
> >samePropertyAs.  These two subrelationships then have no interesting extra
> >meaning, as once two things are the same, they can't be made any samer.
> >
> >So my suggestion is to do away with equivalentTo.
> >
> >Comments?
> 
> Well, an alternative is to do away with all its subrelationships and 
> just use equivalentTo for everything, including classes and 
> properties. We could leave sameFooAs forms there simply as sugar and 
> they would do no harm. Which I think amounts to a proposal to leave 
> it the way it is :-)
> 
> Pat

The subrelationships are there in order to clarify the semantics:
i.e., for <sameClassAs,?C,?D> we have IC(?C) = IC(?D). If ?C and ?D
can be arbitrary then it is difficult to give a clear semantics to
equivalence, e.g., if ?C ?D can contain a mixture of classes,
properties etc. Rather than trying to "disallow" strange types of
equivalence axioms we just went for the standard solution of providing
the generic but semantics free "equivalentTo" along with
subrelationships for classes, properties etc., to which we do give a
semantics. 

We did discuss this extensively before agreeing on the above
"solution" (in fact this dates back to the Dec2000 release) and I am
not sure why we are now revisiting it.

Ian


This archive was generated by hypermail 2.1.4 : 04/02/02 EST