Re: Amendment to KIF's unary-relation and binary-relation

sowa <sowa@turing.pacss.binghamton.edu>
Date: Thu, 26 Aug 93 08:49:44 EDT
From: sowa <sowa@turing.pacss.binghamton.edu>
Message-id: <9308261249.AA07188@turing.pacss.binghamton.edu>
To: interlingua@ISI.EDU, macgregor@ISI.EDU
Subject: Re:  Amendment to KIF's unary-relation and binary-relation
Cc: sowa@turing.pacss.binghamton.edu
I would like to second Bob MacGregor's concern about empty relations.

Suppose someone happens to define a relation with an axiom that
is self-contradictory.  That may be a mistake, but it is a semantic
mistake that a compiler may not be able to detect.

There are also reasons for wanting relations that may be empty in
one possible world, but not in others -- UNICORN(x) for example.
And in the type lattice, there is the universal type at the top
that applies to everything and the absurd type at the bottom that
has no instances in any possible world.  You need the absurd type
if you want to make it a lattice, but the corresponding relation
would have no instances.

John Sowa