Replies: Combining Ontologies (Fritz Lehmann)
Date: Sat, 4 Jun 94 07:40:55 CDT
From: (Fritz Lehmann)
Message-id: <>
Subject: Replies: Combining Ontologies
Precedence: bulk

     About a month ago I asked the attached question about
automated combining of "ontologies" to elicit responses from the
Usenet "" newsgroup readership.  I attach below slightly
edited copies of all the replies I got.

     Additionally, the UMLS Metathesaurus for medical concepts has
been an exercise in "ontology combination" within its field;
I don't know whether or to what degree it has been
"automated" rather than done by hand using human judgement.
There has been sporadic research in comparative library-
classification, as covered by ISKO (Int. Soc. for Knowledge
Organization, Woogstrasse 36a, Frankfurt, Germany).  One
interesting study is M. Dienes' integration of four classification
systems (Unesco Thesaurus, MISON Rubricator [Russian], Universal
Decimal System, and Broad System of Ordering) sharing the common
subject-matter "culture", a nebulous and variously-viewed subject
bravely attacked [M. Dienes, "Structural Differences in
Classification Systems and the Testing of the Compatability Matrix
in the field of Culture", in "Universal Classification II", INDEKS
Verlag, Frankfurt, 1983, p.112].  I imagine that the CYC people
would have come up with something on "microtheory combination" by
now in the wake of Guha's PhD thesis, but I don't really know.
Tony Cohn and I have something called EGG/YOLK theory to deal with
integrating imperfectly matched non-primitive types in differing
ontologies, and Navathe, Convent and others have developed
relevant tools for view integration of database schemata.  Useful
ideas exist in Pawlak's Rough Set theory, Wille's Formal Concept
Analysis, and Russian Classification/Meronomy theory of Shreider,
Polyakov, Kononov et al..

     I believe that there are many further possibilities for
automated combining of concept-systems that have _some_ common
ground, and that the current (apparent) dearth of ideas is

                           Yours truly,  Fritz Lehmann
GRANDAI SOFTWARE, 4282 Sandburg Way, Irvine, CA 92715, U.S.A.
Tel:(714)-733-0566  Fax:(714)-733-0506
>     The following question appeared on the usenet
>newsgroup "" recently:
> >From: windy gambetta <>
> >Subject: combining ontologies
> >Organization: Uni of New South Wales
> >Date: Tue, 29 Mar 1994 23:11:09 GMT

> >I need some info about methods to combine ontologies.
> >could anyone give me pointer to the area ?
> >thanks
>      I didn't see much in the way of answers on usenet.
> How accurate would the following answer be?
>      "At present there are no automated methods
>       to combine ontologies, no formal theories
>       for combining ontologies, and no tools available
>       for combining ontologies except for hierarchy-
>       browsing assistants."
>      Does anyone have counterexamples to this answer?
>                          Yours truly,   Fritz Lehmann

From: ontek!tony@uunet.UU.NET (Anthony K. Sarris)
Subject: Re: Combining Ontologies


[. . .]

If you REALLY understood the question to mean automated combining of
ontologies, then I agree that there is nothing out there, mainly because
there isn't much if anything in the way of formal systems to encode (read
'represent') the semantics of the ontology, and so there's not much to do
automated combining of! In a much more limited sense, I see KIF as being a
way to at least interchange whatever knowledge encodings there are, but I
don't think it does much in the way of automated combining of them, and
clearly humans are needed for the delicate handwork of interpretation and
synthesis. But I guess if two ontologies both 'hosted' their knowledge
encodings in some basic form in KIF, then at least there could be some
level of interchange. From a slightly different perspective, there are
reasoniong systems that might be usable to some degree to support the
combining operation. For example, Logistica from Dr. Frank Brown has some
fairly sophisticated pattern-matching capabilities. To be useful, the
ontologies would have to be encoded in LISP form. Since KIF is fairly
LISP-like, there might be some ways to use the two together. We have
explored re-hosting some of Logistica's algorithms in 'hat form' (Ontek's
syntax), but we don't have an automated link between the LISP structures of
Logistica and the PACIS K-Rep system either.

That's my two-cents worth.

Tony Sarris

Ontek Corporation
22941 Mill Creek Drive
Laguna Hills, CA, USA 92653
Phone: 714/768-0301
Fax: 714/768-0851
Date: Wed, 04 May 1994 15:16:42 +0500
From: presnik@caesar.East.Sun.COM (Philip Resnik - Sun Microsystems Labs BOS)
Subject: RE: Combining Ontologies

[. . .]

A counterexample is work done by Kevin Knight at USC/Information Sciences
Institute; he's merged a number of ontological resources using (semi-
methods.  See Kevin Knight and Steve K. Luk, "Building a Large-Scale
Knowledge Base for Machine Translation," to appear at AAAI-94.  I'd be
interested in hearing if you come across anything else useful.

Hope this helps!



  Philip Resnik                           E-mail:
  Sun Microsystems Laboratories, Inc.     Work:   (508) 442-0841
  Mailstop UCHL03-207                     Fax:    (508) 250-5067
  Two Elizabeth Drive
  Chelmsford, MA 01824-4195 USA
Date: Wed, 27 Apr 94 15:25:44 +0200
From: (Paul van der Vet)
Message-Id: <>
Subject: Combining ontologies

In, you posted the following question:

     "At present there are no automated methods
      to combine ontologies, no formal theories
      for combining ontologies, and no tools available
      for combining ontologies except for hierarchy-
      browsing assistants."

     Does anyone have counterexamples to this answer?

I don't, and I would be very surprised if someone had. Perhaps I'm
mistaken, but I see no flaws in the following argument. Please
correct me if I'm wrong.

In Gruber's enlightening explanation of what an ontology is (KR'92, if
I remember correctly, and an internal KSL report), the definitions
making up an ontology are comprised of a formal and an informal part,
and both are indispensable. You design a formal structure but will
always need primitives whose meaning is explained by anchoring them
into domain knowledge (for example). This can only be done by a
natural language account. Demanding that those meanings, too, are
pinned down by formal definitions would only lead to a regress,
because for those definitions you would need new primitives. And so

Now an automated system able to combine ontologies has to determine
whether the concepts of the two ontologies can be mapped onto each
other. To be able to do that, the system has to reason about the
meanings of the primitives. One way to do that, of course, is by using
a pre-determined third ontology covering both - but that wasn't the
point, was it.


Paul van der Vet                   Phone +31 53 89 36 94 / 36 90
Knowledge-Based Systems Group      Fax   +31 53 33 96 05
Dept. of Computer Science          Email
University of Twente
P.O. Box 217
7500 AE  Enschede
The Netherlands
Date: Sun, 15 May 1994 12:59:25 -0800
To: (Paul van der Vet),
From: Tom Gruber <gruber@HPP.Stanford.EDU>
Subject: Re: Combining ontologies

Hello Paul and Fritz,

I'm going over backlogged mail and found this interested comment that I had
saved.   Myu response, for what it is worth, is to agree that no formal
machinery will be able to completely automate the "combination" of
ontologies that contain primitive (incomplete or informal) definitions.  I
take "combination" to be something like merging the definitions with
consistent renaming and the construction of axioms that map terms in one
ontology to terms in the other.  Formal machinery can help, but will be
incomplete when there are primitives.

This doesn't mean there can be no tools to help.  Just as there are tools
that help humans do natural language translation, there can be tools that
help humans translate and merge ontologies.

By the way, this is the sort of discussion that ought to be going on in the
srkb list (

Feel free to forward your discussion there, or offer permission and I will
forward this message at least.


[. . .]