Re: Some thoughts on K.I.F.
schubert@cs.rochester.edu
Date: Wed, 8 May 91 13:18:18 EDT
From: schubert@cs.rochester.edu
Message-id: <9105081718.AA17073@ash.cs.rochester.edu>
To: INTERLINGUA@isi.edu, SRKB@isi.edu, sowa@watson.ibm.com
Subject: Re: Some thoughts on K.I.F.
John makes some good points. To fill him in a bit, let me mention
a couple of pertinent things that the interlingua group has talked
about. One is a set of "tough nuts", including examples involving
generalized quantifiers like "most". Peter Norvig provided plausible
KIF translations of many of the tough nuts; unquestionably, though,
the translations are rather complex, and wouldn't be easy to translate
back out of -- I personally would favour having a general quantifier
syntax (Q <var> <restriction> <body>), instead of the current KIF
syntax (which for some reason has an arbitrary TERM instead of a
variable restriction. Note: It's easy to supply a first-order semantics
for quantifiers with restrictions.
The second point is that I've proposed a "substitutional definition"
facility for the interlingua. This is similar to a macro facility
EXCEPT that defined symbols can be used directly -- they are not
(necessarily) expanded. In fact, I used the definition of "there
exists exactly one" as one of my illustrations. The specification
of this as part of the interlingua document is in progress. This
definitional facility will make it possible to effectively enrich
the interlingua syntax to make it easier to translate out of.
-Len