Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: SUO: Metaphysical choices - position. mereology and constituti on

Oh John,

For once I almost totally disagree with you, as well as with the rest
of the bunch (which is more usual).

Quoting you from:

> Re metaphysical choices:  The kinds of circles that Chris and Ian
> have been cycling around the question of roles, Queen Elizabeth,
> etc., illustrate my point that many fundamental philosophical
> issues must reach some stage of closure before it is possible to
> write useful axioms that could seriously be considered candidates
> for standardization.

Yes, of course "fundamental philosophical issues must reach some stage of 
closure" before the "cycling" can stop. But I see this more as a discussion
about *content* than about *structure* of the ontology specification.
Why should any choice be favored by the standard? 3d versus 4D etc...
EVERY such statement should be expressible by the "standard".

And I don't think the fault comes from choosing "axioms" too early.
On the contrary I believe the "cycling" comes from the simple fact 
that the arguments of everyone in this debate are NOT formalised enough.
This is the way the philosophy works, the premisses are never fully defined,
not even to say *formalised*, so any line of reasoning can be diverted and 
tweaked according to the current mood and "intuitions" of the various 
participants and the whole thing can never have any "convergence" or 
"confluence" (to be more technical).
This happens to be called "dialectics" which in most (but not all) cases is 
just a well received word for "lies and cheating", but I am digressing...
On one hand this is VERY good, because it allows for creativity, exploring
new domains, reconsidering established (and hidebound) theories, not missing
critical questions, etc, etc...
BUT, if the design of a "common ontology" or (preferably in my view) of
ontologies interoperability is an ENGINEERING project and not just a
philosophic or scientific one, as Bill Andersen ceaselessly and rightfully
remind us, this is NOT the way to go.
Such questions have been debated for centuries, if not millenia and 
you expect to settle those matters with the SUO standard?
You must be joking, or may be you are just plain nuts or idiots?

> Following are the implications:
>   1. Until these issues are resolved, any axioms based on or related
>      to them could never be considered "standard".


>   2. Since roles, intentions, and related issues are fundamental to
>      every topic outside of pure mathematics, that means that any
>      axioms in SUMO that apply to anything outside of mathematics are
>      unsuitable for anything that might be called a "standard ontology".

TOTALLY AGREE on this one!

>   3. Point #2 does not imply that the axioms of SUMO or OpenCyc should
>      be rejected.  It only implies that the axioms should not be given
>      the status of "IEEE standard".  To adopt Frank's suggestion, I
>      believe that a suitable term would be "registered"; i.e., they
>      have been assigned a place and a unique identifier in the kind of
>      registry that Frank was discussing.

I believe this is more a matter of misuse of the word "axiom".
Factual knowledledge encoded by *some* author should not be given the status
of an "axiom" or anything close. It is just just *data* and should bear
some stamp or tag identifying it's source.
And this applies EVEN MORE to philosophical/ontological choices.

>   4. A registry for modules (i.e., collections of axioms) would not
>      imply that any particular module had reached a definitive status
>      that could be called "standard".  It would just mean that it had
>      been registered and agreed to by some number of users for some
>      stated purposes.

And, of course, different users involved in the same field may have
different opinions, purposes and philosophical/ontological premisses.
Should the "standard" favor any one of them?
Should not they be allowed *some* capacity of interaction in spite of
their diverging choices?
Think about it!

>   5. Associated with the modules in the registry could be comments and
>      certifications of compatibility with other modules and various
>      other standards.  For example, a module that axiomatized concepts
>      related to grains and cereals might be certified as compatible with
>      the ISO standard for durum wheat.

NO, plain no, compatibility of "neighbouring" ontologies should not be 
a matter of certification. 

1 - Notwithstanding any "due process" this will be ARBITRARY.
2 - No certification body or process will be responsive enough to the
    needs and speed of the current technology-world and this will get worse.
3 - It will NOT guarantee proper interoperability because it will be
    subject to BUGS like any other software or data entry.

Compatibility of neighbouring ontologies should be AUTOMATICALLY derived 
from a common underlying description (a meta-level) of each ontology
and it is upon the shape and meaning of this meta-level that an agreement
must be reached. To get back to the 3D/4D dilemma again, the right question 
is not which is the best for *most* purposes or even how to make both
live side-by-side but how to find a *common* description scheme that
will allow to encode both AND map them back and forth as much as possible.

Meanwhile any discussion about which is best as well as about "queenness 
of Elizabeth Windsor" or "purpose" versus "intended purpose" etc, is just
IRRELEVANT (dare I say bullshit, yes, I do).

> Bottom line:  I would strongly object to any application of the term
> "IEEE Standard" to any collection of axioms related to the topics that
> Chris and Ian were discussing.  But I would be quite happy with putting
> them into a registry, where the users could certify their applicability
> and compatibility with other related standards.

Still objecting to the word "axiom" (could be "rule" or whatever), I agree
at least that refering to some identifier from a common registry to specify
upon which premisses the semantics of some knowledge set is to be build is 
the very minimal thing to do.
But again the question arise, what is the encoding of the registry contents?
I maintain that it must TOTALLY formalised, this is why I support things
like IFF, but, on the other hand, I DO NOT believe that such complete
formalism can capture natural languages or common sense!
And this is why I strongly object about any attempt to solve 
questions *today* about natural languages or common sense, 
this is science not ENGINEERING.

But, there is a LOT to do before we really have to deal with natural 
languages or common sense. That will be less fun for the philosophers 
but of more PRACTICAL import, see Bill's message:


-- Jean-Luc Delatre
"The mind of man is more intuitive than logical, and comprehends
 more than it can coordinate."         -- Vauvenargues, 1746
--------------------------------------------------------------------------  -- GSM: +33 6 11 24 06 29