Some thoughts about one of Matthew Wests's comments: (01)
[MW] >> The current direction seems to be
one of a merged lowest common denominator approach. I think that to be
successful, we will have to move to a lattice of theories approach, as
described by John Sowa. (02)
"lowest common denominator" would not be a proper characterization of
the suggested common subset. "Maximum common denominator" would be
closer, but even that would underestimate what might be achieved by an
effort at interrelation. A "lattice of theories" might be useful to
relate logically incompatible ontologies within a single framework, but
may be beyond what can, as a practical matter given the likely level of
funding, be achieved in the near future. The goal does not require
that the format of the common subset be identical to that of the linked
upper ontologies, only that some accurate translation can be created
from the simpler common ontology to the more expressive upper
ontologies; "common subset" may be a misleading term, but was used in
place of an explanatory paragraph. (03)
A question that the panel might consider worth pursuing is how large
and expressive a "common subset" could be found (1) without any changes
in the existing upper ontologies; and (2) with small changes in one or
more of the existing upper ontologies that can be accommodated within
their dominant paradigms of representation. Discovering these
relations would be one possible project agreed to by the panel, to be
pursued subsequent to the meeting. I believe that the technical
barriers are not large to creating such a common subset ontology,
provided that the upper ontology custodians think that the goal is
sufficiently worthwhile that it would justify making some small
modifications of their upper ontologies to increase the breadth and
expressiveness of the common subset. (04)
One immediate need that makes this meeting timely is the proliferation
of models (e.g. in UML) and ontologies, e.g. in OWL, that have limited
capability to express detailed semantics for the relations. It could
be very helpful to have wide agreement on even a relatively simple
axiomatic ontology that provides some detailed semantics for the most
commonly referenced entities, and relations, e.g. people,
organizations, events, artifacts, dates, locations, parts, causality.
At present, references are being made to Dublin Core and Friend of
Friend, without any careful definition of what the referenced entities
mean in an inferencing environment. A common subset that provides at
least the minimum axiomatization to eliminate the major potential
ambiguities would, I believe, provide a significant service to the
growing community who want to explore the potential of ontologies. (05)
The larger opportunity that such a common subset might present is to
provide a gentle introduction for those who are already creating
knowledge representations in simple formats to the need and opportunity
to create representations with more detailed and more computationally
useful meanings. As potential users learn more of the capabilities of
logic-based ontologies and become more comfortable with the axiomatic
specification of intended meaning, that should increase interest in the
more fully functional upper ontologies, leading to an increase in the
number of their users. (06)
Pat (07)
Patrick Cassidy
MITRE Corporation
260 Industrial Way
Eatontown, NJ 07724
Mail Stop: MNJE
Phone: 732-578-6340
Cell: 908-565-4053
Fax: 732-578-6012
Email: pcassidy@xxxxxxxxx (08)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/uos-convene/
To Post: mailto:uos-convene@xxxxxxxxxxxxxxxx
Community Portal: http://ontolog.cim3.net/
Shared Files: http://ontolog.cim3.net/file/work/UpperOntologySummit/uos-convene/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?UpperOntologySummit (09)
|