I've been trying to stay out of this, but it seems to me that the model
proposed to me by Subrahmanian Eswaran at CMU (and I've never seen the paper,
if he published it) is appropriate here. (01)
Sub's model says that in any discipline there is a "core of accepted
terminology" that develops over time. The terminology initially appears in a
book or paper, and then a few books and papers, and then in industry
discussions, and then in academic course materials, etc. They all begin as
"folksonomies" emanating from a single source or a small group. (02)
Acceptance of the prevailing core of common terminology becomes a part of
being understood in conversations with one's peers in the discipline. You
find out that a term is on the fringe when someone whose expertise you respect
is confused by the term. But there is no clearly fixed boundary to the core
terminology for the discipline. Acceptance is a fuzzy value. (03)
Over time, new concepts take hold through the same introduction mechanisms,
and they often require refinements to earlier concepts, because the universe
of that discipline has gotten bigger by their very existence. And these
ripples take a while to propagate across the users of the terms. So the
"folksonomy" of the discipline is constantly in flux, but the amplitude of the
variance is small. Even the core is not immune to change, but the core is
recognizable by a much longer cycle of change: 25-50 years. (04)
Now all of this presumes that you are talking about a discipline in which
publication occurs and communication is a frequent activity of the
participants. In such areas, we can be comfortable in formalizing what we
perceive as the accepted core. Anything beyond that is based on our relative
comfort with some estimated value of "acceptedness" that is less than 1, but
still "statistically significant". (05)
Some areas, and in particular some in which we are being asked to develop
ontologies, do not have identifiable disciplines! Others are locally
controlled silos, in which communication outside the fence is just not very
common, or actively discouraged. In those areas, there is no commonly
accepted core of terms -- many concepts may be very closely aligned, but it is
hard to find enough common starting points to make that judgement. (06)
Even when there is a robust discipline, that doesn't necessarily mean that the
terminology is disciplined. Many terms are used by different practitioners
with slightly broader or narrower meanings. It is characteristic of
folksonomic concepts that there is wide agreement on the central
characteristics of a concept, but little or no agreement on exact necessary
and sufficient characteristics. Part of this problem is the ripple effect
mentioned above. And part of the problem is what Pat Hayes called the
"Horatio Principle" (there are more things in heaven and earth than are dreamt
of in your philosophy): not every expert has seen, or cares about, the
examples that violate his would-be classification razors. (07)
In hard natural sciences and a handful of other disciplines, the reference
terminologies and taxonomies exist, there is a strong terminological
discipline, concepts are always carefully defined and revised when necessary.
For those, we can safely build reliable ontologies. (08)
Almost any other ontology we may develop is necessarily artificial -- a
significant part of it is NOT based on accepted terms with accepted
definitions. In effect, it is an artifice based on a "folksonomy", and its
utility is directly proportional to its "acceptance coefficient". (09)
I think most business communications, most government communications, and many
professional communications are based on "folksonomies" whose underlying
disciplines, or at least their terminological disciplines, are weak. I think
we can make useful ontologies for those, but the applications have to be very
narrow, until and unless Sub's "core of accepted terminology" becomes exact
enough and large enough to support many applications. (010)
-Ed (011)
--
Edward J. Barkmeyer Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263 FAX: +1 301-975-4694 (012)
"The opinions expressed above do not reflect consensus of NIST,
and have not been reviewed by any Government authority." (013)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (014)
|