SUO: Re: (SeWeb) engineering ontologies?
Michael Sevcenko wrote:
> And this brings me to the second question: is it permissible
> to use KIF for construction of ontologies, or it is obsolete
> due to DAML+OIL?
KIF and conceptual graphs are much richer languages than
DAML+OIL. They are intended for a much wider range of
applications than DAML+OIL. See the current work on the
Common Logic Standard:
http://cl.tamu.edu
The main purpose of DAML+OIL is to use a simple subset of
logic that is efficient for classifying terms. However, if
you want to give a complete definition of the terms (i.e.,
to write axioms that define their meaning and use), you will
need the richer languages.
The PSL and SUMO ontologies, for examples, use full power
of first-order logic as expressed in KIF. The Cyc and
OpenCyc systems use the CycL language, which is approximately
of the same expressive power as KIF and CGS (but the actual
work to determine the exact mappings remains to be done).
Those languages go beyond FOL to add the ability to represent
contexts and metalevels.
The intent of the Common Logic standard is to define a common
model-theoretic semantics, which would ensure that rich
languages, such as KIF, CGs, and traditional predicate
calculus can freely exchange information among themselves.
We hope that CycL can brought into the same group of languages,
but that work remains to be done.
For more restricted subsets, such as DAML+OIL, it is possible
to transfer information into the richer languages, but the
richer information cannot not be translated back to a more
restricted language.
John Sowa