Wednesday, March 01, 2006
Generative Methodology Glass Bead Games
On the limits of the OWL standard à [184]
Reading material [1]
Reading material [2]
Reading material [3]
Summary of the discussion up to this point à [186]
On ontological modeling of biological expression
Subtitle: regarding the 'expressiveness" of OWL in the
case that there are emergent natural categories... how does one create the
proper evolution from an individual to a class?
On Formal verses Natural systems à [206]
On the BioPAX forum we had the following communication
Hi
Frank -
You
stated that OWL ...
>
...seems to have problems describing the structure of a
>
relationship of three or more items to each other.
That
is a popular myth, but quite untrue.
The trick is to
use
OWL design patterns involving structures of triples.
http://biopaxwiki.org/cgi-bin/moin.cgi/OntologyDesignPatterns
best
regards,
Dan
Communication from Paul Prueitt
The issue of restrictions on the n-ary relationship is only part of the problem and I agree with Gary that W3C disclaimers would help everyone understand what OWL can and cannot do.
At the beginning of the recommendation it is stated
"The document presents ontology patterns for representing n-ary relations and discusses what users must consider when choosing these patterns. "
It would be nice to have an objective examination of why a "class" restriction is not applied to a property. Is this basically about the barrier in description logics to allowing a property to become an class?
John Sowa, is this the issue?
I will argue that the W3C recommendation moves in the direction of allowing an property to become an individual. One has to do this manually, I assume.
But the W3C standards cannot allow that individual to become an class. But I may have missed something.
But let me see if I understand the recommendation at:
http://www.w3.org/TR/swbp-n-aryRelations
and ask the authors of this recommendation if I am correctly representing the recommendation.
The standard states
"A common solution to representing n-ary relations such as these is to create an individual which stands for an instance of the relation and relates the things that are involved in that instance of the relation. We can think of the original relation then as a class of all these relation instances."
But, in my understanding, a class is thought of as the set of individuals that reify the class !!
This feature of description logic allows the "inference" engine to return an individual that satisfies the definition of membership in a class if and only if the properties of the class match exactly to that individual. The exact match might not be explicit, thus the notion of "inference".
In fact this "inference" is "all" of the mechanism involved in created explicit representations of what is implicit in the OWL indexing. Again correct me if I am wrong.
(I may have gotten this a bit wrong.)
The technical problem regarding treating individuals as classes seems non-trivial to solve to the non-logician, but the logician is required to do a type of shell game, there the completeness and consistent conflict plays out. Right?
I conjecture that the specific issue faced by the BioPAX "physicalEntityParticipant" class, is that description logics ARE NOT OPEN to the conversion of instances to classes.
This is why some scientists call the so-called open world assumption of DL, a partial open world assumption.
(I am really guessing on some of this, so please correct me where I am wrong.)
The natural world may have a feature whereby an instance (something that actually occurs) modifies the natural categories related to it's set of entailments. So the natural category modification must be reflected in the class structure of any ontological model "of the biology".
Can OWL be used to achieve this openness of categorical modification in the vivo.
http://en.wikipedia.org/wiki/In_vivo
This is an open question for me.