Dear Matthew, Pat, Pat, and Ali,
I agree with that point:
MW> I don't think Longman's will do much good for us, though I
equally doubt it will do much harm. Frankly, I'd rather PatC
got on with doing something with it so we had some evidence
we could look at rather than debate a priori what the effect
We've all seen many failures of committees to reach a consensus.
Furthermore, we've seen a growing number of independently
developed ontologies. Most of them pick and choose excerpts from
one another. But they never converge on a common foundation.
MW> There inevitably are/will be several upper/foundation ontologies.
I do think there is a chance to abstract something as John says
underspecified, and I do think that could be useful. This is not
really what PatC is after, but it is not incompatible with his
I think we need to allow multiple upper ontologies and mappings
between them. Underspecified/abstract elements would be useful
to make the mapping easier. They do take on different meanings
when added to different ontologies, but that is fine.
Yes. The hierarchy of theories is designed to accept any reasonable
contributions of any size. It can then make all interrelationships
among the ontologies and subontologies explicit. By itself, the
hierarchy is unbiased and egalitarian. But the reviews by users and
domain experts can provide the guidance for making rational choices.
PH> The other worry I have about 'intended interpretation' is that,
even when working alone, one finds that the very process of writing
the formal axioms sharpens and sometimes forces one to modify ones
own pre-formal intuitions.
Yes. And the users also modify their intuitions and intentions
when they see the results. Engineers have a slogan:
Customers never know what they want until they see what they get.
PC> But part of the problem is mitigated by allowing multiple different
ways of representing the same entity in the FO, as long as they are
logically compatible and have translations between them....
The need to refine intuitions and record those distinction in
logically precise form has always been part of the ontology-building
process. Hard work to be sure, but not impossible.
That statement can serve as a good basis for collaboration.
AH> In fact, except for the quest to identify and create a special
foundation ontology out of the primitives, what you are now
proposing is indistinguishable from what is already underway via
COLORE. COLORE is gathering all ontologies written using CLIF
regardless of their terminology or quirks. It provides a growing
platform (a repository) in which they may be inputted and relations
between them explored, identified and formalized.
As we have agreed in other email notes, the COLORE methodology
is completely compatible with the hierarchy of theories. The work
that has already been done for COLORE is an excellent beginning.
But I would add some further extensions to the COLORE slides:
1. The ontologies can be represented in any notation that has
a formal mapping to Common Logic. That includes all the
Semantic Web ontologies and many others. But that mapping
must be done before they can be admitted to the hierarchy.
2. Various projects and their developers and users have different
preferences and requirements for notations. Therefore, any
ontology in the hierarchy may have multiple representations
for each axiom, and the tools for displaying and editing
ontologies can have options for highlighting or suppressing
some of the options.
3. UML diagrams and controlled natural languages (CNLs) have also
been translated to Common Logic, and many users have found them
to be much more readable than CLIF and other notations used
for logic. The SUMO developers, for example, use controlled
English for comments that can be automatically translated to KIF.
Such notations should not only be supported, they should be
4. Many important ontologies that could be converted to CL have
not yet been converted. SUMO, for example, was written in KIF,
which is a predecessor to CLIF. Most statements in KIF can be
converted to CLIF with little or no change. But a few rarely
used features of KIF must be checked for compatibility.
5. The CycL language used for Cyc can mostly be converted to CL
with little or no change. But the full CycL language uses
metalevel features that require the IKL extensions to CL.
We should consider broadening the scope to include IKL.
6. Some of the diagrams in those that show relationships among
theories can be enhanced or redrawn to show the relationships
to the lattice of theories.
7. The limitation of OOR to just the notations currently
supported by Protege is untenable. OOR must support full
CL and even IKL. The Protege tools should be extended
or combined with tools that support those languages.
Point #7 is critical. The Semantic Web includes an important
collection of technologies, which must be supported. But they
have already moved beyond RDFS and OWL by including Horn-clause
logics. Furthermore, relational databases still support the
world economy, and the constraints and queries used for RDBs
are written in full first-order logic. The EXPRESS language,
which Matthew and others have been using, also requires full
FOL. And Cyc, the biggest formal ontology on earth, uses a
superset of FOL. The OOR *must* support these systems.
I am currently writing a report based on excerpts from recent
email notes to Ontolog Forum that elaborate these points.
I'll post a first draft by this weekend. Next week, we can
discuss further modifications and updates on Ontolog Forum,
and I'll post a new version of the report by March 12th.
After the Ontology Summit at NIST on March 16th, we can
discuss that report and other related issues in the session
scheduled from 3:45 to 5:15.
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx