Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

SUO: 12 May 2002 -- Procedural Questions




¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤

Procedural Questions

AP = Adam Pease
BA = Bill Andersen

cf.  http://suo.ieee.org/email/msg08441.html

BA: What properties do all SUO entries have to have?

AP: Comprehensiveness, internal consistency, clarity, economy --
    all motherhood sorts of things :-)

BA: What are the concrete testable criteria for judging when we
    have something sufficient to fill the role of a standard?
    In other words, where's the finish line?

AP: While it may be unsatisfying, most software has an open ended
    charter and deciding when it is "done" is a somewhat arbitrary
    choice.  We have several indications of impending completeness
    on SUMO:

AP: 1.  The rate of change of the ontology is slowing significantly.

AP: 2.  We've mapped over 50,000 WordNet terms to SUMO and found that most
        of what isn't directly present in SUMO is of a very specific nature.

AP: It's always possible to release a version 2.0 of any software once bugs
    or enhancements are agreed upon.  No software, and no standard will ever
    be "perfect", so one uses tests such as these to make a reasonable choice
    as to when to release.

cf.  http://suo.ieee.org/email/msg08447.html

AP: Creation of criteria for ontology "goodness" I suspect would be even harder
    than creation of an ontology.  Plenty of good products, including standards,
    are created without a priori precise quality criteria.  Our criteria, such
    as they are, already exist and are embodied in the current PAR.

Most of what AP say is nonsense, of course.  There are plenty of
external criteria as to "what is standard elsewhere" (WISE) that
we might draw on as measures of ontological "fitness" to purpose.
So it is not a matter of "creating" our own criteria for grading
ourselves, so much as reflecting on existing criteria and trying
to render them explicit, for our own good.

Nothing about this desideratum need inhibit innovation.
It is simply the pre-condition of a genuine innovation.

To say that SUMO must be approaching a state of "good enoughness"
on the evidence of narcissistic and solipsistic internal criteria,
or the fact that its rate of progress is retarding, is just plain
silly, given that its comport with what is standard elsewhere in
so many areas -- set theory, graph theory, formal language theory,
algebra, topology and real analysis, pure and applied mathematics
in general, the theory of computation and programming languages,
information theory, probability and statistics, systems theory,
semiotics, ..., just to mention a few that I happen to know of --
is nowhere near being up to par with an average undergraduate
major's competence in these subject areas.  The SUMO document
in its present condition is a work of impressive mathematical
illiteracy and a major fund of misinformation on almost every
subject it touches -- I would not have thought it possible to
remain so totally innocent of basic acquaintance with so many
of these standard topics in these times, if I had not actually
seen it in print.

Perhaps the SUMO Team thinks that it has some secret plan for
flying under the radar of expert critique and external review
on all of these basic measures of "goodness", but I think that
it would do them more good to meditate on the possibility that
this would not really be such a "good thing" for anybody, even
if they could.

Jon Awbrey

¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤~~~~~~~~~¤