SUO: Re: Standard Upper Ontology Procedures
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
SUOP. Note 2
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o
[Reposting after 1.5 hours]
JA = Jon Awbrey
JS = John Sowa
John,
I no longer see much likelihood that this working group,
or another gang of "French Encyclopedists", will be able
to develop a Standard Ontology, upper or otherwise. It's
been 3.5 years of banging my keyboard and my head against
walls, both concrete and abstract, and here are some of
the impressions that have been formed, all on my head:
1. Some ontologies are innate concept systems.
2. Some ontologies are inured belief systems.
3. Some ontologies are testable scientific theories.
4. Some ontologies are test-immune religious beliefs.
All of these categories of ontological complexes are interesting from
one disciplinary perspective or another, and a Standards Body may have
some part to play in classifying them according to their characteristics,
but the kinds of critiquable, deliverable, explicitly formulated ontology
projects that a Standards Body can do anything about critiquing, testing,
and recommending improvements in are largely found behind Door Number 3.
A few years back, a number of participants were suggesting
that a big part of our preliminary work would have to be
the "Ontology Of Ontologies" (or "OOO" to me). I do not
know who first introduced the term, but an initial search
of the archives reveals that some of the people talking
about this were Fred Chase, Doug McDavid, Leonid Ototsky,
Sofia Pinto, and Mike Uschold. Although I thought this
idea was interesting, from a metareflective perspective
or "strange loop" point of view, I don't think that
I got the critical importance of it for our work
at that time.
After all this time, I cannot think of a single instance where
any ontology, personal or starter-documented, has changed in
any significant and substantive way during the course of
our deliberations. That recalcitrant fact demands that
I look for a different model of what is feasible here.
1. In spite of how much I wish it were possible,
I do not think that we are currently engaged
in "open-source ontology development" (OSOD).
The conditions for doing that sort of thing
are just not in force at the present time.
2. I now believe that we really are in the position of a very
standard sort of "standards and practices" body, like some
sort of federal highway safety admininstration, or whoever
it is that evaluates the saftey of roads, tires, vehicles.
We are in the position of having different tire companies,
say F-year and G-stone, dump a load of already-been-made
tires on our doorstep, subject to evaluation and testing.
The properties of these various lots of tires are just not
about to change. They just lay there. Questions about how
they were made can be asked, but will be tied up in court for
years, protected by all the usual brands of proprietary dodges.
We may seek to shift the workload, or avoid the responsibility,
by asking companies to submit their self-evaluative test data.
But, as Matthew is fond of saying, we and all our loved ones
deserve everything that happens to us if we fail to inquire
carefully into that data. (As I rather less fondly observe,
it's usually X that suffers the fate that Y deserves, where
X =/= Y. For example, X = Major Tom, Y = Ground Control.)
Bottom line? It appears that a major component of the function that is
relegated to this Standards Body is to classify "tires" (in the analogy,
ontology documents) according to their testable and tested properties,
to put stickers on them that say things analogous to "suitable for use
as a backyard planter", "okay to drive on if it doesn't rain", "unsafe
at any speed", "AOK", and so on.
Jon Awbrey
JS: I agree with Jon on the following two points:
JA: We already have lots of procedures and
routines for voting on starter documents.
I suggest that we not haggle over those
sorts of issues anymore, indeed, I have
come to revise my own opinion about what
qualifies as a starter document to what
already seems to be the de facto norm,
that just about any working effort or
historically established project is
worth discussing in this forum.
JS: I believe that any serious work on any subject related
to ontology should be seriously considered as a potential
contribution to the SUO effort.
JA, correcting awkward constructions and subject-verb agreements in JA:
JA: It seems that one of our major roadblocks at this time has to do
with the very different sorts of ontology projects before us, and
though I believe that this diversity can be a good thing, I believe
that it raises several questions to ask about each of these projects:
JA: What does this particular ontology project want to be? How does
it want to define itself, its aims, its own criteria for success,
and how does that project for being relate to the other projects?
JS: Those are questions that have not been adequately answered,
and the directions taken by some (or perhaps all) of the
current starter candidates do not cover the full range
of what should be done.
JS: I would recommend the following book about the development
of the Oxford English Dictionary as something to consider:
| Simon Winchester,
|'The Meaning of Everything,
| Oxford University Press, 2003.
JS: This is a fascinating account of the development of the OED from
its proposal in 1869 to its completion in 1928 -- a period of 59
years. The time from the first publication of the first fascicle
in 1882 to the final fascicle in 1928 was 46 years.
JS: That time period should be compared to the 19 years for the
development of Cyc, which is very far from satisfying anybody
who is looking for an ontology that could be considered worthy
of being a "standard".
JS: I do not see any reason for assuming that the
development of a good SUO would take much less
time than the development of the OED.
o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o~~~~~~~~~o