Re: Good and Bad IS-A hierarchies (Pat Hayes)
Message-id: <>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Wed, 26 Jul 1995 14:42:14 -0500
To: Peter Clark <>, (Doug Lenat)
From: (Pat Hayes)
Subject: Re: Good and Bad IS-A hierarchies
Precedence: bulk

>True too: axiom 1 (containers need portals) makes an ontological commitment 
>about containers which precludes osmosis. We've been looking a bit at this:
>we can "localize" this commitment by introducing a specialized container, 
>("portaled-container", say), and then asserting this axiom just for that 
>specialization, rather than all containers. Now we can introduce osmosis into
>our representation somewhere else without conflict. 

Well, it depends what you mean....One way here is to categorize container
types, as you suggest. (A problem which tends to arise is that one gets isa
heirarchies with unnecessary depth due to odd cases which need to be
classified off the edge: osmotic containers would be a godd example, in
fact.) But another way is to keep your axiom as it is, and say that in the
case of osmosis the entire surface of the container is portal (of a rather
special kind). Then all the action is in characterising what kinds of sheet
or surface can be an osmotic portal for what kinds of stuff. This seems to
be the way one thinks in considering vapor barriers and even insulation
barriers (heat as a kind of liquid..) in house construction, for example,
where a "solid" wall can be more like a bug screen for water vapor.
(Actually, by the way, osmosis is more complicated, involving differential
"vapor" pressures and walls which seem permeable in one direction but not
the other.) 

I rather prefer this approach because it keeps secure the very basic
intuition that in order to get from inside to outside, anything has to
somehow pass through the boundary; and also because it focusses on surface
types, which seems to be important. But the real issue is, suppose you do
it your way and I do it my way: how will we then be able to communicate?
"portal" wont mean quite the same to you as it does to me, for example. 

In fact -- an "isa" 
>hierarchy falls out naturally by looking at what things there are in the 
>world, identifying which axioms are potentially applicable, and then looking 
>for subset/superset relationships between those axiom sets. The (general) 
>axioms are the building blocks, and the taxonomy reflects the things which 
>you can build with them, composed out of those building blocks but using
>specialized names (eg. "container" -> "portaled-container") to localize the
>ontological commitment. 

I offer the axiom collections in my 'temporal catalog' as a challenge to
any such endeavor.

Pat Hayes

(If anyone's interested we've a rather half-baked 
>working note at with a bit 
>more on composing action descriptions from components).
>                Best wishes,
>                        Pete
>Peter Clark (   Department of Computer Science
>tel: (512) 471-9565                  University of Texas at Austin
>fax: (512) 471-8885                  Austin, Texas, 78712, USA.
>Project homepage:

Until September: 
1916 Ivy Lane, Palo Alto, Ca. 94303               phone (415)855 9043