[Top] [All Lists]

Re: [ontac-forum] Future directions for ontologies and terminologies

To: ONTAC-WG General Discussion <ontac-forum@xxxxxxxxxxxxxx>
Cc: 'SUO WG' <standard-upper-ontology@xxxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Fri, 30 Dec 2005 14:20:09 -0500
Message-id: <43B58869.8080408@xxxxxxxxxxx>
Azamat, Lucian, Arun, and Pat,    (01)

I'm replying to your comments in this forum, but I'm
cc'ing this note to the SUO list.  I suggest that the
more detailed discussions be continued in that forum.    (02)

AA> Your persistent criticism of a unified framework
 > ontology (UFO) is counting more on rhetorical points,
 > metaphors, analogies, and figures than on substantial
 > grounds and facts.    (03)

My recommendation, which I've mentioned many times before,
is to take advantage of two major resources:  the 11179
standard for metadata to register the implementations and
the infinite lattice to use as an abstract framework:    (04)

  1. I am not against the concept of a UFO, and it may be
     possible that an infinite Supreme Being may have such
     a vision in His, Her, or Its mind's eye.    (05)

  2. But every proposal that any human has put forward so
     far has been fatally flawed, and I have addressed the
     arguments for such UFOs in the same manner in which
     they have been proposed.  I don't believe *any* of the
     pro arguments have been able to withstand the criticism,
     but I leave that question for the users to decide.    (06)

  3. In any case, my recommendation is *not* to abandon the
     search for a single UFO, but to recognize that the search
     is far from complete and that the terminologies are far
     more stable, complete, and reliable than *any* detailed
     formal axiomatizations of those terms.    (07)

  4. Therefore, my proposed solution is to establish a metadata
     registry of any and all ontology modules (which may be
     complete or partial ontologies at the upper, middle, or
     lower levels) and to define systematic methods (which can
     be supported by practical tools) for combining and relating
     those theories to one another and combining them as needed.
     (The operators discussed in my theories.htm paper are an
     example, but a lot more can be added to those 13 pages.)    (08)

  5. Such a registry could be established right now according to
     the 11179 standard, everybody who has their own ontology
     (complete or partial) can register it tomorrow, and users
     can begin using them and recording their experiences as
     comments added to the registry.    (09)

This approach provides an immediate interim solution, which can be
adopted now, and the accumulated comments can be used as a basis
for selecting the use cases necessary to make informed decisions
about the future.  Instead of basing a decision on endless email
wrangling, users can look at the comments in order to choose the
module(s) that have proved to be most successful for applications
similar to their own.    (010)

AA> To be serious, the major obstacle to UFO is that the high
 > task demands profound scholarly learning, intellectual
 > dedication and consecration to fundamental study.    (011)

I certainly agree, and I would add a great deal of experience
with practical implementations and extensive testing.    (012)

Meanwhile, people need something they can use today for critical
applications.  My greatest fear is that the pressure to provide
an immediate solution will cause people to jump onto the first
available bandwagon, such as the crowd that adopted QDOS (Quick
and Dirty OS, which Bill Gates bought for $50,000 in 1981 and
sold to IBM for one million dollars as MS-DOS 1.0).    (013)

AA> The good and useful practice of SUO demonstrates that we
 > must follow some basic methodological principles and strategic
 > assumptions to construct a unifying foundational ontology.
 > Among them, I believe, are the following...    (014)

Your points and Arun's additions are well taken.  My major
addition is the need for practical experience, extensive
testing, and most importantly a systematic mechanism for
revising and correcting the inevitable errors and shortcomings.    (015)

LR> One of the decisions one needs to make about an upper Ontology
 > is whether it is needed at all.  If the answer is “yes” then
 > the choice is to create a single upper Ontology that categorizes
 > all human knowledge into a small number of categories. This is
 > by definition static. The other alternative is to have a large
 > number of categories which can be combined on the fly; this is
 > by definition dynamic. This decision entails recognizing the
 > form it would take.    (016)

I agree, and I also believe that we do not yet have sufficient
information to make a firm commitment to either alternative.
That is the motivation for the above proposal, which supports
multiple intermediate ontologies that can be registered and
used *as if* they were standards, but which provides methods
for relating them to one another within a theoretical framework.    (017)

LR> Some researchers in Artificial Intelligence claim they have
 > a better top level Ontology, and maybe they are justified.
 > However, WordNet has some 15 years behind it, so the set-
 > descriptions of the any different concepts should be carefully
 > spelled out so that we may judge.    (018)

WordNet is certainly the world's most widely used resource for
ontology/terminology, and most people who are developing similar
resources have adopted, adapted, modified, used, or considered
it in one form or another.    (019)

PC> Could you illustrate what you mean by combining categories
 > "on the fly"?  I think that we all agree that any common upper
 > ontology will have to grow, evolve, and be supplemented over
 > time;  Does this nevertheless qualify as "static"?    (020)

I can't answer for Lucian, but I would like to use the example
of a metadata registry as a repository of implementations and
the infinite lattice of theories as the abstract framework.
Each version of each implementation would be registered and
would remain unchanged (i.e., static) in the repository.  Each
new version would be assigned a new entry in the repository,
and the metadata would specify exactly how it was derived from
and related to the older version(s) and/or any other resources.    (021)

Meanwhile, the abstract lattice provides the framework, which
can be navigated by means of the four theory revision operators.
That navigation path from one abstract theory to another could
be called a dynamic or "on the fly" revision of the theory at
the starting point of the path.  For a discussion of the theory
revision operators, see Section 5 of the paper:    (022)

    http://www.jfsowa.com/logic/theories.htm    (023)

One of the major influences on my proposal above is that almost
everybody (including Cyc, SUMO, etc.) who has developed a highly
axiomatized ontology has deleted the WN upper levels and replaced
them with their own.  Then they aligned the lower levels of the
WN synsets with their own categories.    (024)

The widespread use and success of that approach is one of my
reasons for arguing that a resource, such as WN which emphasizes
terminology and minimizes detailed axioms, provides a good
unifying focus.  Many people who have used or studied WN have
criticized it for one reason or another, and I would not recommend
it as an official standard.  But WN certainly deserves a place
of honor in a registry along the lines proposed above.    (025)

John Sowa    (026)

Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki: 
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG    (027)
<Prev in Thread] Current Thread [Next in Thread>