Re: SUO: Starter Ontology Version 2
> I agree that a "basic semantics for the most general structural
> features of the language" might be useful, and it would probably not have
> the adverse effect of prejudging substantive issues. However, if I
> understand what you mean by the term, a "basic semantics" would also not
> have the crucial property of telling us whether Mr. Whitten's axioms (or any
> other axiom set formulated within SUO KIF) is consistent. I have no
> objection to the development of a "basic semantics", if, again,
> I understand you, but I also don't see that it would have much of a bite.
Well, there's bite and then there's bite. Of course, insofar as the basic
semantics only fixes the core structure of the ontology -- issues of
typing, the logical status of properties and relations -- and gets object
language and metalanguage issues sorted out, it will not provide an
interpretation of the particular constants and predicates of a more
detailed ontology. Okay, no bite in that sense. It doesn't give you the
consistency of particular ontologies for free. However, once those issues
*are* fixed, and object language / metalanguage stuff sorted out, I
believe it will in most cases be straightforward to verify the consistency
of most ontologies by constructing models of those ontologies that extend
the basic semantics directly. Indeed, often, I think, this will be quite
simple -- basically just getting the logical connections between classes
right. But even in more complex cases the basic semantics will provide a
rigorous, easily extensible foundation to build upon that will enable
model construction to proceed in a standard, uniform way. From this
perspective, the bite of a basic semantics is, I think, deep.
I believe it is a mistake to lump David's commendable work together with
"any other axiom set formulated within SUO KIF". The chief problems that
I saw in his work are just the sorts of problems that I believe the
existence of a basic semantics (which David did not have the benefit of
consulting) will help ontology authors to avoid in the future. Notably,
as it stands, in addition to reflecting the lack of resolution of the core
issues of typing and the like, David's ontology also contains a
substantial amount of detailed metatheory, including a number of semantic
notions -- notions that we know pose a signficant threat of inconsistency.
(Please note I am not *faulting* David for this, it is just an issue that
needs to be dealt with that his work has helped to highlight.) These
problems largely go away, I believe, if the core structure of the ontology
is fixed by a basic semantics and the metalanguage / object language
issues are sorted out.
Regards,
-chris
--
Christopher Menzel # web: philebus.tamu.edu/~cmenzel
Philosophy, Texas A&M University # net: chris.menzel@tamu.edu
College Station, TX 77843-4237 # vox: (979) 845-8764