Re: Ontological EDI & Wittgenstein
phayes@cs.uiuc.edu (Pat Hayes)
Message-id: <199409301940.AA25446@dante.cs.uiuc.edu>
X-Sender: phayes@dante.cs.uiuc.edu
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Fri, 30 Sep 1994 14:42:26 +0000
To: scottmd@cts.com (Scott M. Dickson), fritz@rodin.wustl.edu (Fritz Lehmann)
From: phayes@cs.uiuc.edu (Pat Hayes)
Subject: Re: Ontological EDI & Wittgenstein
Cc: cg@cs.umn.edu, edi-new@tegsun.harvard.edu, srkb@cs.umbc.edu
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk
At 2:22 PM 9/29/94 -0800, Scott M. Dickson wrote:
>Fritz Lehman wrote:
>>
>> Yeah, I thought about possible exceptions to the mutual
>>exclusion of BARGE and SUBURB; that's why I said "we can agree
>>completely that no barge ... is a suburb" rather than "No
>>barge can be a suburb."
>
>Or, we might not agree. The question is, do we have to?
Yes!! That is, we (and our programs) have to somehow come to sufficiently
close terms of agreement in order to be able to communicate properly.(Or be
able to understand one another's vocabulary; at any rate we need to get to
sufficiently 'deep' mutual understanding that each can translate the
other's meaning into their terminology.)
>A deeper taxonomy would recognize that anything that can contain or carry
>something could be used as a unit of measure, although in English we often
>add the suffix -load or -full to indicate this as in truckload, shipload or
>capfull.
Actually I dont think this is 'deep' enough. Its not the carrier that is
the unit, but the amount of stuff that the carrier holds. (See T. R. Gruber
& G. R. Olsen. An ontology for engineering mathematics. In Jon Doyle, Piero
Torasso, & Erik Sandewall, Ed., Fourth International Conference on
Principles of Knowledge Representation and Reasoning , Gustav Stresemann
Institut, Bonn, Germany, Morgan Kaufmann, 1994.) Caps are not amounts of
liquid.
This is a nice example of the kind of distinction that English is often
careless about, which might seem unnecessarily 'philosphical' to some, but
which might be crucial to keeping mechanical inference straight. For
example, if any container IS a unit, then it follows that I can divide my
Ford Festiva by an integer. Moreover, any solid thing could presumably be
used as a unit of weight, and then it follows that (since my cup is both a
unit of volume and a unit of mass) that there are things that have
dimension m and dimension l!3; so all dimensional reasoning turns into
styrofoam; etc etc. As soon as you get a contrdiction in the system all
hell breaks loose. Now of course these silly conclusions don't bother your
average chap, and seem ludicrous. They ARE ludicrous, and thats the point.
Not because of the logic, but because of carelessness in descring the
situation in the first place; carelessness which we all committ when
speaking to one another in English without even noticing that it is
happening.
Maybe there are robust forms of reasoning that can somehow not be bothered
by such contradictions and inconsistencies, but its not enough to just
notice that people can do it. We know that. What we dont know is HOW people
do it, and how we can make machines not do it.
And (in case anyone agrees that its time to unsubscribe because all this is
irrelevant...) as soon as you allow communication between systems, this is
a centrally practical issue. There are systems out there that do
dimensional reasoning. If your system ever communicates ideas with one of
them, it had better not tell it anything that allows it to conclude that a
mass is a volume.
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
----------------------------------------------------------------------------