Re: Ontologies (Pat Hayes)
Message-id: <>
X-Sender: (Unverified)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Mon, 24 Jul 1995 16:48:47 +0000
To: (John Thompson)
From: (Pat Hayes)
Subject: Re: Ontologies
Precedence: bulk
John Thompson wrote:

>Toward the end of his long and interesting letter Patrick Cassidy
>>...  Even with
>>the top 2000 concepts agreed upon, there will still be enormous 
>>scope for proprietary details and variations in the programs that 
>>will use the ontology.
>>   To be really standard, an ontology should also include 
>>some level of agreement on semantic relations (slots) that are
>>used in defining terms.  Without some commonality in relational
>>links, it will be hard to verify that the meanings of the
>>individual nodes in a semantic network have the same
>>significance to all systems.
>Yes, and the semantic relations (slots) should be organized as part of
>the ontology itself.  In the Cyc ontology, if you go down about 6 or 7
>levels you come to the collections Predicate and Slot (binary
>predicate), and under those (as instances) are the actual predicates,
>arranged in their own ontology. 

All this amounts only to agreeing on the vocabulary. But what really
matters is agreeing on the axioms which use the vocabulary. For example, we
can agree to use the concepts of 'timepoint', the binary relation 'before'
between them, the concept 'timeinterval' and the functions 'begin' and
'end' from timeintervals to timepoints. We can agree to all this and even
to a number of plausible axioms connecting them all together, and yet still
have starkly contradictory views on whether, for example, intervals can be
said to "meet" one another, and whether the usual distinctions between open
or closed intervals make sense. Can something be both an interval and a
point? Can the timeline branch into alternative futures? None of these can
be addressed by agreeing on vocabulary.

More significantly, however, is the fact that we might disagree on
vocabulary and still agree on essential content, since different
vocabularies are sometimes easily intertranslatable. For example, you might
prefer to have the trinary relation 'meets-at' between two intervals and a
point, while I might prefer to only talk of timepoints being 'contained-in'
timeintervals and think of the interval as a kind of approximate
specification of the timepoint. With some reasonable axioms on each side,
these vocabularies are easily intertranslatable. James Allen might want
thirteen relations in his vocabulary: thats fine, since we can define them
all in terms of 'meets-at' in our theory. (But if someone wants to describe
an interval as a set of points, then we have some genuine problems...)

The point is that all this fussing over is-a heirarchies and getting
vocabularies straight really misses the core issue, which is locating the
key conceptual clusters and trying to get enough of the content axiomatized
in a useful way. Once this is done, it is easy to be flexible about
particular vocabularies.

Pat Hayes

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