WD-Ontology

[Sebastien Derriere]

Ed Shaya wrote:
> 
> There are dangers in defining the necessary and sufficient conditions,
> particularly if one requires too many necessaries.  You can end up with
> things that are not in any class although they are so close to one that
> ordinarily one would include them.  

  Hi Ed,

  Of course it is irrelevant to be over-constraining in the definition
of necessary and sufficient conditions. We certainly don't want to
incorporate
in ONE definition too many constraints: giving the definition of "Star"
as
X and Y and Z and "hasMeasurement some EffectiveTemperature" is too
restrictive because it would forbid an object to be a star simply if 
no effective temperature has been measured, even when X, Y and Z are
satisfied!
  But I insist that necessary and sufficient are very important, because
they are the basic building block for reasoning. We can easily overcome
the problem given in my example by having multiple definitions (sets of 
necessary and sufficient conditions, each restricting to a minimum) for
a
same concept. I think we both know this, but I want to make it clear for
non-specialists.
  Also in our ontology we try to place additional constraints by:
* setting disjonction between concepts (Star disjoint from Galaxy)
* adding "reverse" conditions via Inverse properties. e.g. we don't say
Star: "hasMeasurement some EffectiveTemperature", but we constrain the
inverse property of hasMeasurement, isMeasuredFor, by saying:
EffectiveTemperature: "isMeasuredFor only Star"

  (I give here a simplified example, see the owl file for actual usage)
  The reasoner can use these informations, and we're not too
restrictive...

> For instance, what if it is like a
> cataclysmicVariable but not quite a LateType star. One also run into
> whether or not to use strictly observational qualities or interpreted
> qualities.  As in, what if it erupts and has emissions like a
> cataclysmic variable but it is too far away to discern the individual
> components?  Can you still call it a CV?

  Again, you could give one set of NSC based on observables, and one
on interpreted qualities for a same concept.
  And because the subsumption relation has multiple levels, even if you
can't  infer an exact match, you can in most cases find that concepts
are
"close" because they have the same direct subsuming concept.

> On the other hand, if we don't make clear and distinct boundaries you
> are left with a system that is too fuzzy to do any good.

  Agreed, the reasoner only deals with "true" or "false". It can't
understand astronomer's fuzzy logics like "well, mostly true, I guess,
but...".
Hopefully there is already a lot to constrain with exactly true or
false statements for the ontology to be useful.

Sebastien.

-- 
    _______
   /  ~   /, Sebastien Derriere   mailto:derriere at astro.u-strasbg.fr
  / ~~~~ //  Observatoire de Strasbourg    Phone +33 (0) 390 242 444
 /______//   11, rue de l'universite     Telefax +33 (0) 390 242 417
(______(/    F-67000 Strasbourg  France