The bio-ontologists currently developing the phenotype and other
ontologies have addressed this question in detail. Their earlier
attempts led to what I propose calling the 'Peanut Butter Sandwich'
problem. If you have Attributes (Color) and Values (Red), then for
some attributes (e.g. Height) you will infinitely many values, and
for other attributes (e.g. Eats) you will have uncontrollably many
made-up values (e.g. Peanut Butter Sandwich with McDonald's Chicken
Fajita Sauce and a Pickle). (01)
The solution they are working on is to drop the whole notion of
Values. Rather, there are determinable attributes (Color) and
determinate attributes (Red). Attributes are not relations between
bearers and values. Rather, every single attribute instance, for
instance the color of Rudolf's nose, instantiates a series of
attribute types at greater and lesser levels of granularity. (02)
This still leaves open the problem of Height. Here the solution is
along the lines of accepting Height as a determinable attribute, with (03)
Height-of-2-Meters
Height-of-1.9-Meters (04)
etc. (05)
as determinates. In any given domain of biological inquiry, there is
a finite number of such relevant determinates.
BS (06)
At 03:30 AM 1/21/2006, you wrote:
>COSMO-WG:
> To follow Leo's suggestion:
>
> One technical issue we need to resolve immediately is the
>representation of attribute types and attribute values (Color ->
>RedColor; Length -> 34 cm; Apprehensiveness -> Terrified), because
>these differ among the different starting ontologies. In fact, they
>differ between OpenCyc 0.7 and OpenCyc 0.9. They also differ and
>present problems in implementation, when function terms are used as the
>representation for measurement values. Each of the representations is
>coherent, we just have to choose, and there are several issue involved
>in choosing.
>
> Anyone want to get into this? Send me a note, and we will form a
>CosmoTaskTeam on the issue. If it goes over 10 discussants, perhaps it
>should stay on the dev list.
>
> Pat
>
>
>Patrick Cassidy
>MITRE Corporation
>260 Industrial Way
>Eatontown, NJ 07724
>Mail Stop: MNJE
>Phone: 732-578-6340
>Cell: 908-565-4053
>Fax: 732-578-6012
>Email: pcassidy@xxxxxxxxx
>
>
>-----Original Message-----
>From: ontac-dev-bounces@xxxxxxxxxxxxxx
>[mailto:ontac-dev-bounces@xxxxxxxxxxxxxx] On Behalf Of Obrst, Leo J.
>Sent: Friday, January 20, 2006 5:41 PM
>To: ONTAC Taxonomy-Ontology Development Discussion; Chris Menzel
>Subject: RE: [ontac-dev] Type vs. Class -- Please vote
>
>All,
>
>Personally, I think a better methodology for these problematic
>terminology aspects is to choose the disjunction:
>
>Class OR Type
>
>And move on.
>
>Here's a more general suggestion:
>
>Have small teams of perhaps 3-5 people be designated to address
>Problem, where Problem is a variable that ranges over the problematic
>issues we try to determine or at least list as a group (so at a
>meta-level). These small teams will go away and analyze the Problem,
>and report back with a Strawman Recommendation. We NEED to make
>progress and all this very interesting thrashing won't get us anywhere.
>
>
>Here's some preliminary suggestions (in no particular order, because
>some depend on others, and are necessarily incomplete):
>
>1) Universal/Generic characterization: Class OR Type (or if you are not
>happy with that: Extensional vs. Intensional)
>2) PartOf characterization; Constitution characterization.
>3) Properties, Relations, Attributes, Tropes, etc.
>4) 3D vs. 4D or both
>5) Entity vs. Process (state, event, activity, action, etc.)
>Time/change stuff.
>6) Necessary and sufficient characterization of all of the above
>7) Your favorite other stuff
>...
>
>
>_____________________________________________
>Dr. Leo Obrst The MITRE Corporation, Information Semantics
>lobrst@xxxxxxxxx Center for Innovative Computing & Informatics
>Voice: 703-983-6770 7515 Colshire Drive, M/S H305
>Fax: 703-983-1379 McLean, VA 22102-7508, USA
>
>
>-----Original Message-----
>From: ontac-dev-bounces@xxxxxxxxxxxxxx
>[mailto:ontac-dev-bounces@xxxxxxxxxxxxxx] On Behalf Of Cory Casanave
>Sent: Friday, January 20, 2006 3:18 PM
>To: 'Chris Menzel'; 'ONTAC Taxonomy-Ontology Development Discussion'
>Subject: RE: [ontac-dev] Type vs. Class -- Please vote
>
>Ok, logical gurus, help me out here.
>The asserted "difference", between type and class is extensional. The
>statements below say that if the instances satisfying the types are the
>same
>then types are the same.
>Another statement would be if the instances satisfying the types must
>be the
>same the types must be the same.
>The <are> and <must be> are very important.
>
>The first statement would only work for predicates that assume a very
>closed
>and very static world. E.G. the type of people on this list and the
>type of
>people with 874 hairs on their left arm are the same so these types
>must be
>the same - a very bad inference, particularly after I pluck a hair. It
>assumes global knowledge and a static world. While I have no problem
>that
>predicates or even logics may choose to make those assumptions, it
>should
>not be part of the definition of type (or class).
>If, on the other hand there is a logical necessity for the extensions
>to be
>the same, the class and type concepts are the same - this works and I
>don't
>see the issue. This also works in OWL. This is what I thought the
>concept
>of extensional classes entailed - but I am still learning.
>As Chris says, you can ADD those assumptions to an ontology or context
>without constraining core concepts like type/class/isKindOf.
>
>
>-----Original Message-----
>From: ontac-dev-bounces@xxxxxxxxxxxxxx
>[mailto:ontac-dev-bounces@xxxxxxxxxxxxxx] On Behalf Of Chris Menzel
>Sent: Friday, January 20, 2006 12:33 PM
>To: ONTAC Taxonomy-Ontology Development Discussion
>Subject: Re: [ontac-dev] Type vs. Class -- Please vote
>
>On Fri, Jan 20, 2006 at 06:27:49PM +0200, Christopher Spottiswoode
>wrote:
> > ...
>
>Hello, Christopher, it's been a while.
>
> > Pat, my own vote goes strongly to "type" (if I may emerge albeit
> > perhaps incongruously from my lurking state on this list).
> > Im my own work, which I hope to release to this list and the widest
> > user community at a more appropriate time, I have for many years
> > consistently and insistently used "type" to denote the intensional
> > sense. "Class", it seems to me (as to John Sowa), invites confusion
> > with the extensional sense.
>
>I continue to be utterly flummoxed by this argument. Do we, or do we
>not, believe ontologies can avert exactly these sorts of confusions?
>If not, just what do we think we are doing? Looky here:
>
>(forall (C1 C2)
> (if (and (Class C1)
> (Class C2)
> (forall (x)
> (iff (instanceOf x C1)
> (instanceOf x C2))))
> (= C1 C2)))
>
>Add that axiom to your ontology, you get extensional classes. Leave it
>out, you don't. Simple, eh? :-) What *is* the controversy here?
>
> > Having said that, I would however agree with Cory that we should try
> > to conform as far as possible with what seems like colloquial use.
> > But I think that that rather argues in favour of "type", as it is (to
> > me at least...) more colloquially intensional than "class"!
>
>Well, at the risk of furthering the impression that this red herring of
>an argument is to the point :-) , I respectfully disagree. Indeed, I
>can't even think of a colloquial use of the term "class" that is
>extensional. Not even "set" is extensional in colloquial usage.
>
> > I mention that because both Chris Menzel and Leo Obrst have warned us
> > against using "type" because of all the uses of that word in various
> > formal systems.
>
>That skews my point badly. It is not simply the fact that "class" is
>the term of choice in the formal system OWL that we should use it; it
>is
>the fact that it is in OWL AND the fact that OWL and its kin are the
>primary W3C languages for publishing ontologies on the web. We're
>swimming unnecessarily, indeed perversely, upstream if we choose
>otherwise. Seems to me that the only thing that could justify the
>choice of "type" would some definite semantic incompatibility between
>the desired ONTAC notion and the W3C notion of class. But there isn't.
>So if we go with "type", we force EVERY user of OWL out there who wants
>to interact with an ONTAC-based ontology needlessly to worry about
>translating every occurrence of "type" into "class". Similarly for
>every user of any ONTAC-based ontology. Isn't the point here to
>*enhance* interoperability? Why throw up this completely unnecessary
>stumblingblock, folks?
>
> > I would strongly urge us, however, not to be so influenced by such
> > perhaps confusing formal uses:
>
>It is the informal uses that are confused. We have axioms to prevent
>that sort of confusion.
>
>-chris
>
>
>_________________________________________________________________
>Message Archives: http://colab.cim3.net/forum/ontac-dev/
>To Post: mailto:ontac-dev@xxxxxxxxxxxxxx
>Subscribe/Unsubscribe/Config:
>http://colab.cim3.net/mailman/listinfo/ontac-dev/
>Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
>Community Wiki:
>http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatin
>gWG
>
>
>_________________________________________________________________
>Message Archives: http://colab.cim3.net/forum/ontac-dev/
>To Post: mailto:ontac-dev@xxxxxxxxxxxxxx
>Subscribe/Unsubscribe/Config:
>http://colab.cim3.net/mailman/listinfo/ontac-dev/
>Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
>Community Wiki:
>http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatin
>gWG
>
>_________________________________________________________________
>Message Archives: http://colab.cim3.net/forum/ontac-dev/
>To Post: mailto:ontac-dev@xxxxxxxxxxxxxx
>Subscribe/Unsubscribe/Config:
>http://colab.cim3.net/mailman/listinfo/ontac-dev/
>Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
>Community Wiki:
>http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatin
>gWG
>
>_________________________________________________________________
>Message Archives: http://colab.cim3.net/forum/ontac-dev/
>To Post: mailto:ontac-dev@xxxxxxxxxxxxxx
>Subscribe/Unsubscribe/Config:
>http://colab.cim3.net/mailman/listinfo/ontac-dev/
>Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
>Community Wiki:
>http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG (07)
_________________________________________________________________
Message Archives: http://colab.cim3.net/forum/ontac-dev/
To Post: mailto:ontac-dev@xxxxxxxxxxxxxx
Subscribe/Unsubscribe/Config: http://colab.cim3.net/mailman/listinfo/ontac-dev/
Shared Files: http://colab.cim3.net/file/work/SICoP/ontac/
Community Wiki:
http://colab.cim3.net/cgi-bin/wiki.pl?SICoP/OntologyTaxonomyCoordinatingWG (08)
|