Common ontologies
Patrick Cassidy <micra@tigger.jvnc.net>
From: Patrick Cassidy <micra@tigger.jvnc.net>
Message-id: <199506100232.AA15142@tigger.jvnc.net>
Subject: Common ontologies
To: cg@cs.umn.edu
Date: Fri, 9 Jun 95 22:32:23 EDT
Cc: srkb@algol.cs.umbc.edu
X-Mailer: ELM [version 2.3 PL11]
Sender: owner-srkb@cs.umbc.edu
Precedence: bulk
The recent expressions of agreement on the utility of
finding some common ground in development of ontologies prompts
a suggestion about how to proceed.
(1) each group willing to try to find common ground in building
ontologies should provide each other group a copy of whatever
parts of their system could eventually be released, either
(a) into the public domain, or (b) minimally restricted.
Proprietary considerations can be important, so I will make
a suggestion on that point as well:
The common ontology to be developed should be available
via ftp and other public distribution channels to promote
rapid exchange of ideas, as well as comments from individuals
not necessarily participating actively in the development of
the ontology. If only the top 1000 or 2000 concepts are to
be made common, these should be made public domain, but if that
would inhibit participation by some groups, the ontology may
still be copyrighted, with the copyright held jointly by the
participating groups. Since we expect only a small part of each
group's system to be made public in this way, this should not pose
a serious risk to those with heavy financial investments at stake.
Even if copyrighted, a publicly available "open" status seems
to me to be important.
(2) each group should examine the other groups' systems,
including the hierarchy, the predicates required for definitions
of these top categories (semantic relations, frame slots), and
the mechanism of inheritance (multiple? defeasible?), with a
view to finding some common areas where their systems are similar,
or can be easily translated by machine.
(3) from each group's own perspective, they should try to
outline a common ontology which would take into account the
maximum areas of agreement among the different systems.
(4) The resulting set of possible "common" ontologies can form
the starting point for the discussion of the feasibility of finding
a single common system that a majority can agree on.
(5) Groups that need formal internal discussions to decide
whether to participate in this way should try to have a
decision by the time of the August Montreal workshop.
If not attending that workshop, they could communicate their
decision to me. I will see that the message is relayed to the
workshop participants. A decision to contribute data or
participate in discussions will of course *not* constitute
a commitment to adopt any system developed.
(6) I hope that any who think that this approach is unfeasible
will suggest a different route to the goal.
PatC