In developing a unified framework, we need to get all
the major players in the ontology field to work together
right from the beginning. (01)
Since all the major systems are currently incompatible
with one another, that requirement imposes constraints
on what is possible. Therefore, I propose the following
*neutrality principle*: (02)
The unified framework UF should be neutral with respect
to all the major ontology projects that are currently
under development. That implies: (03)
1. Every system X that participates in the effort
should support import and export operators for
importing all of UF or any subset of UF to and
from X. (04)
2. UF should not contain any categories or relations
that would create an inconsistency with any major
system X; i.e., it should be possible to import
*all* of UF into X without causing an inconsistency. (05)
3. Importing UF into any system X and then exporting
it from X should result in a version UF' that is
logically equivalent to the original UF except for
possible cosmetic changes in the formatting. Those
changes should not cause any other system Y that
imported UF' to generate inferences that differed
from the inferences generated directly from UF. (06)
4. Points #2 and #3 imply that the initial version of UF
should avoid having a complex or detailed upper level,
since most of the inconsistencies between any two
ontologies result from problems at the top. It also
implies that the system should contain a minimal
number of relations whose definitions are not overly
restrictive; i.e., it is better to have *too few*
axioms than too many, since the more axioms there
are, the more conflicts arise. (07)
5. Point #3 implies that the emphasis of the UF should
not be on rich inference capabilities, since those are
usually highly context dependent and very likely to
lead to inconsistencies. Therefore UF would be better
suited to interchange and communication than to extended
inference or problem solving. The extended inferences
would be done by more specialized systems, which could
add additional axioms of their own and use either
logic-based methods or computational techniques. (08)
6. UF should avoid features that limit its use to any
particular notation or system of inference. OWL,
for example, could be used to represent all of UF,
but UF should not have any dependencies on any features
of OWL -- either in logic or in formatting -- that are
not available in all major systems of ontology. (09)
The details of these points are negotiable, but the fundamental
principle of neutrality should be that UF shall be based on the
minimal subset of features that do not create inconsistencies
with any major ontology. (010)
To avoid slighting anybody, I'll avoid listing what ontologies
should be considered "major". (011)
John Sowa (012)
_________________________________________________________________
Message Archives: http://colab.cim3.net/forum/ontac-forum/
To Post: mailto:ontac-forum@xxxxxxxxxxxxxx
Subscribe/Unsubscribe/Config:
http://colab.cim3.net/mailman/listinfo/ontac-forum/
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki:
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG (013)
|