Re: Knowledge Representation Problem
phayes@cs.uiuc.edu (Pat Hayes)
Message-id: <199409282142.AA21459@dante.cs.uiuc.edu>
X-Sender: phayes@dante.cs.uiuc.edu
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Wed, 28 Sep 1994 16:44:23 +0000
To: dwig@markv.com, cg@cs.umn.edu
From: phayes@cs.uiuc.edu (Pat Hayes)
Subject: Re: Knowledge Representation Problem
Cc: sowa@ranch.poly.edu, cg@cs.umn.edu, interlingua@ISI.EDU, srkb@cs.umbc.edu
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk
At 12:17 PM 9/28/94 -0700, dwig@markv.com wrote:
>Pat Hayes writes:
>> Hi John
>
>> Thats a nice problem.
>
>> Is the goal to simply produce a rational conceptual organisation sufficient
>> to help a human programmer implement something, or to formalize the problem
>> and the relevant background assumptions in enough detail to be able to draw
>> some valid conclusions about it?
>
>> Pat
>
>I can't help responding to that "simply".
Im sorry, I didnt mean to imply anything by the "simply", it was only a way
of speaking.
Assuming that the goal is to
>implement something more than just "something", e.g., a reservation system
>that can adequately support the kinds of functions described, the human
>programmer will of necessity have to "formalize the problem and the relevant
>background assumptions in enough detail to be able to draw some valid
>conclusions about it".
Yes, but my distinction was between a human programmer being able to make
those inferences, and an inference engine being able to. For example, a
sufficiently detailed and precise description in English might suffice for
the first, but not for the second. (For that matter, a sufficiently
elaborate and opaque set of axioms might be exactly the other way round.
For example, KIF syntax, intended for machine use, makes things distinctly
less readable for most humans.)
The nature and purposes of the programmer's
>formalizations and conclusions will be rather different than those of the
>philosopher or AI researcher, however. His concerns are more immediate and
>focused on the operational aspects of the problem.
>
The AI researcher might be equally concerned with the immediate operational
aspects. The whole point of using such an example as an exercise in
formalisation is to see if one can do it in a way that would enable the
machine to draw the practically useful conclusions, hopefully in a
reasonable time.
Pat Hayes
----------------------------------------------------------------------------
Beckman Institute (217)244 1616 office
405 North Mathews Avenue (217)328 3947 or (415)855 9043 home
Urbana, IL. 61801 (217)244 8371 fax
Phayes@cs.uiuc.edu
----------------------------------------------------------------------------