"Close enough"

fritz@rodin.wustl.edu (Fritz Lehmann)
Date: Wed, 16 Feb 94 10:48:22 CST
From: fritz@rodin.wustl.edu (Fritz Lehmann)
Message-id: <9402161648.AA09580@rodin.wustl.edu>
To: cg@cs.umn.edu, interlingua@ISI.EDU
Subject: "Close enough"



     Ramesh Patil seems right about harmonizing Bob MacGregor's
practical disregard of KIF's formal semantics with Pat Hayes'
insistence on its importance.  Reserving names for certain things
needed now while awaiting later formalization is a good idea.   But
MacGregor's "close enough" standard seemed to go dangerously beyond
that (and beyond the freedom of "Lehmannism" in the recent
Hayesism/Lehmannism thread) in that he seemed to allow for outright
disobedience of KIF's existing formal semantics.  He said earlier:
"In fact, there is an inverse relationship between how much will get
translated and how tightly coupled the translators are to the KIF
semantics."

     Although some logicians might respond with contempt, it seems
more valuable to demarcate exactly what the problem areas are, and
figure out and define a principled level of "near-compliance" if real
compliance is impossibly burdensome on practical work.  (Maybe non-
empty-intersection could be a substitute for equivalence in
translating certain specified kinds of predicates, for example.  Tony
Cohn and I recently cooked up a theory called the EGG/YOLK lattice
for prototype-based "loose" translations.)

     Since I got on the interlingua email list, there have been few
statements about the nuts-and-bolts practical problems of using KIF. 
MacGregor has mentioned three during that time; I'd like to see a lot
more.  Doyle & Patil's paper complaining about expressiveness
amputations of KL-ONE was based partly on real feedback from
practitioners.  It would be good to see that kind of feedback for KIF
discussed on this list.  People should explain the problems and give
some real examples.  Also it would good to say when Tom Gruber's
Ontolingua "frame ontology" is and is not helpful, and why.


                          Yours truly,   Fritz Lehmann
4282 Sandburg, Irvine, CA 92715  714-733-0566  fritz@rodin.wustl.edu
====================================================================