Re: Browsing Ontologies

Tom Gruber <gruber@HPP.Stanford.EDU>
Message-id: <199408262029.NAA04432@HPP.Stanford.EDU>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Fri, 26 Aug 1994 13:31:48 -0800
To: (Mike Gassmann),
From: Tom Gruber <gruber@HPP.Stanford.EDU>
Subject: Re: Browsing Ontologies
At  2:54 PM 8/26/94 +0200, Mike Gassmann wrote:
>In a workshop we discussed the abbilities of Ontolingua.
>We stopped at the point we want to browse Ontologies, manipulate the database
>and see the effects on the system.

Why stop there?  The ontology library has tons of examples, heavily indexed
for easy browsing on the World Wide Web at
Even the introductory documents are linked into the formal ontologies using

>Are there any design-tools or guides expect the reference-manual to design
>ontologies on my own?

Not enough, that's for sure.  There are some groups working on editing
tools, and I hope they will be at a stage to announce them soon.  I refer
students to the ontologies in the library as exemplars to get started.  One
thing we've learned in the past few years: ontologies need to be designed
for shareability -- it doesn't follow from just doing a good job at
philosophy ("design for truth") or making it work for an application
("design for operationality").

>Are there any papers about it?

I have written two papers on the subject.  One a proposal of some
evaluation criteria for ontology design, with two case studies.  The other
is an in-depth presentation of a familiy of ontologies with discussions of
some engineering strategies for design-for-shareability.

They are also on the web, at

We hope to have some more papers, from around the world, on this subject in
the library soon.

>Are there papers about advantages from Ontolingua vs. e.g. C++ or so?

Ontolingua and C++ are very different beasts; comparing them would be like
comparing an architect's blueprint with a brand of concrete.  Ontolingua is
for _specification_ of ontologies; it can translate the specifications into
some _representation_ languages.  C++ is a programming language that is
especially bad at knowledge representation.

The closest thing to C++ is CLIPS, a C-based production rule system with a
simple object-oriented data structure facility that people have used for
knowledge representation.  Ontolingua can translate KIF into CLIPS, and the
CLIPS files can be loaded into C programs.  There are agents at Stanford
that load in the CLIPS files to get the class hierarchy of ontologies into
the programs.

Thanks for your questions.  I hope that ontolingua can play some part in
specifying knowledge sharing agreements in your project.