Re: Higher-order KIF & CGs

Harold Boley <boley@informatik.uni-kl.de>
Reply-To: cg@cs.umn.edu
Date:     Thu, 4 Nov 93 11:30:33 MET
From: Harold Boley <boley@informatik.uni-kl.de>
To: phayes@cs.uiuc.edu
Cc: Fritz Lehmann <fritz@rodin.wustl.edu>, boley@dfki.uni-kl.de, cg@cs.umn.edu,
        interlingua@isi.edu
Subject:  Re:  Higher-order KIF & CGs
Organization:  University of Kaiserslautern, DFKI, Germany
Message-id: <9311041130.aa19442@uklirb.informatik.uni-kl.de>

Hi,

just some little remarks concerning this final PPS in Pat Hayes' yesterday's
reply to Fritz Lehmann:

> PPS. Your point about the multi-adicity of KIF screwing up some of the
> conventional instantiation schemes is a good one, and I think more work
> needs to be done on this in general. I agree that this is expressively very
> useful and natural, but it has given rise to a number of annoying technical
> problems, and maybe it needs to be looked at more carefully.

I've been working on the formalization of 'variable-arity' data/operators.
E.g.

Boley, H.: A treatment of collection data as constructor algebras.
Univ. Kaiserslautern, FB Informatik, MEMO SEKI-84-06, Oct. 1984.

was an attempt to capture algebraically n-ary tuples, sets, bags,
(hyper)graphs, etc. in the fashion LISP reduces n-ary lists to binary
cons nestings.

In the FIT language, coming from the PLANNER tradition, I used (multiple)
"sequence variables" in arbitrary argument positions (non-deterministic!) of
n-ary relations/functions. In the current RELFUN I only use a (single) rest
parameter (deterministic!) in the LISP sense; while KIF writes something very
similar with general sequence variables, RELFUN sticks to ordinary (rest-
*list*-valued) variables using PROLOG's vertical-bar notation not only within
lists but also in structures and on the top-level of relational/functional
definition patterns and calls. The language also has higher-order sugar for
constructors, relations, and functions:

@inproceedings{Boley92,
   author = {Harold Boley},
   title = {{Extended Logic-plus-Functional Programming}},
   editors = "Eriksson, Lars-Henrik and  Halln{\"a}s, Lars and
              Schroeder-Heister, Peter",
   booktitle = {Workshop on Extensions of Logic Programming, ELP '91,
                Stockholm1991},
   publisher = {Springer},
   series = {LNAI},
   year = {1992}
}

Best regards,

Harold Boley
DFKI
Postfach 2080
D-67608 Kaiserslautern
Germany

Phone: +49-631-205-3459
Fax:   +49-631-205-3210
Email: boley@informatik.uni-kl.de