ontac-dev
[Top] [All Lists]

RE: [ontac-dev] Representation of attributes

To: "ONTAC Taxonomy-Ontology Development Discussion" <ontac-dev@xxxxxxxxxxxxxx>
From: "West, Matthew R SIPC-DFD/321" <matthew.west@xxxxxxxxx>
Date: Tue, 24 Jan 2006 19:21:14 -0000
Message-id: <A94B3B171A49A4448F0CEEB458AA661F02CE52B0@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>
Dear Barry,    (01)

See below.    (02)


Regards    (03)

Matthew West
Reference Data Architecture and Standards Manager
Shell International Petroleum Company Limited
Shell Centre, London SE1 7NA, United Kingdom    (04)

Tel: +44 20 7934 4490 Mobile: +44 7796 336538
Email: matthew.west@xxxxxxxxx
http://www.shell.com
http://www.matthew-west.org.uk/    (05)


> -----Original Message-----
> From: ontac-dev-bounces@xxxxxxxxxxxxxx
> [mailto:ontac-dev-bounces@xxxxxxxxxxxxxx]On Behalf Of Smith, Barry
> Sent: 23 January 2006 09:32
> To: ONTAC Taxonomy-Ontology Development Discussion
> Subject: RE: [ontac-dev] Representation of attributes 
> 
> 
> 
> > > > > 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.
> > > >
> > > >MW: Do you mean here that (your) redness of Rudolphs nose is
> > > an instance
> > > >of a set of varying and usually widening spectrum ranges?
> > > i.e. you are
> > > >dealing with the issue of accuracy/precision?
> > >
> > > Sets do not have instances.
> >
> >MW: Then what please is the name of the relation you use to relate a
> >set to one of its members? How does its nature differ from 
> that between
> >a type and one of its instances?
> 
> membership
> 
> 1. the number 4 is a member of the set {4, the moon, Napoleon}
> 2. there is no corresponding type    (06)

MW: Why? Surely a trivial type listing the members can be constructed.
> 
> 3. Harvey is a member of the set of rabbits
> 4. Harvey is also an instance of the type rabbit.
> It is in virtue of 4. that 3. is true.
> 
> We see that there are cases of set-membership where there is no 
> counterpart to 4. Hence set membership is at least a wider notion.    (07)

MW: Yet membership would not be time indexed whilst instance_of could
be (for you at least). Sounds to me as if each is narrower than some
common core. Can I assume (correctly) you consider that they have the 
same properties in terms of transitivity and the like?
> 
> Sets are abstract entities, with wonderful mathematical properties, 
> but they are not what, e.g., biologists study.    (08)

MW: Interesting how often it is people don't recognise what they are
looking at.
> 
> > > Types have instances.
> > > We represent these instances using various means, e.g. 
> English words
> > > ('red', 'bright red') or hexadecimal numbers, or what you will.
> > > Sometimes our representations are more precise, sometimes 
> less. They
> > > may still all be correct (as it may be equally correct to say:
> > > 'animal over there', or 'cat at fifty paces').
> >
> >MW: I think I have mostly grasped what you mean by a type, surprising
> >as it has been to me. Just one last clarification here. I presume you
> >agree there are some types whose membership does not change, 
> i.e. your
> >type and set have the same members, e.g. integers and real numbers.
> 
> If integer is a type, then the set of its instances is indeed always 
> identical to the set of its instances. And ditto for 'real number'.    (09)

MW: So one possibility here might be for me to settle for types whose
instances didn't change over time, and just forget about the "sets"
as uninteresting.    (010)

MW: My problem now is that I understand that you want types to be
restricted to things like rabbits, not "people with 376 hairs on their
arms" or "4, the moon, and me". Now, whilst I am sympathetic
to the idea of natural kinds, it seems to me that these three examples
actually sit in a spectrum and there is no clear divide between them
(though these three being prototypical can be easily distinguished).
So I am looking for something that is broader that would encompass all
these, within which you could distinguish your interest perhaps as
(someone else suggested) natural types.
> 
> >MW: Do you still insist that the type and the set are 
> different objects
> >in this case? (It would seem somewhat redundant to me to do so, but I
> >suspect that you do insist).
> 
> So much ontological nonsense has been created because people thought 
> crucial distinctions were redundant.    (011)

MW: And as much again from not recognising when things were the same.
> 
> >MW: If we are to find any commonality it seems to me we need 
> to follow
> >Chris Menzels approach where he was saying that 
> class/type/sort/category
> >is not extensionally defined unless you have the axioms to 
> make it so.
> >
> >MW: So I suggest a structure as follows:
> >
> >1. Class/type-that-may-or-may-not-be-defined-extensionally.
> >
> >Some subtypes of this would be:
> >
> >2. class/type-defined-extensionally.
> >
> >3. class/type-defined-intensionally.
> >
> >Since this could include people with 374 hairs on their arm. 
> A subtype
> >of this that would be:
> >
> >4. Class/type/sort-defined-intensionally.
> 
> This, again, is putting the definitional cart before the ontological 
> horse. There were billions of types trillions of years before there 
> were any definitions. Definitions are nice, if you can have them, but 
> types do not depend for their existence on definitions.    (012)

MW: No argument from me here, I was just a little loose with language.
> 
> The above, ontologically, is no better than a taxonomy along 
> the lines of
> 
> class Michael West defined    (013)

MW: Is this supposed to be anyone I know?    (014)

> class Michael West defined before breakfast
> class Michael West defined before breakfast during a lunar eclipse    (015)

MW: Ah well. I withdraw the olive branch.
> 
> >This should at least allow us to identify the very different 
> things we
> >seem to be interested in. We can then start trying to understand the
> >relationship between them.
> 
> 
> > > > >
> > > > > This still leaves open the problem of Height. Here 
> the solution is
> > > > > along the lines of accepting Height as a determinable
> > > attribute, with
> > > > >
> > > > > Height-of-2-Meters
> > > > > Height-of-1.9-Meters
> > > > >
> > > > > etc.
> > > > >
> > > > > as determinates. In any given domain of biological
> > > inquiry, there is
> > > > > a finite number of such relevant determinates.
> > > >
> > > >MW: Are there not an infinite number of heights between
> > > Height-of-1.9-Meters
> > > >and Height-of-1.9-Meters? Or are you saying that we are only
> > > interested in
> > > >the values at 1mm distances between these?
> > > >
> > > >MW: This still seems somewhat simplistic for engineering 
> purposes.
> > > There is infinite complexity everywhere. I am sure your 
> Oil ontology
> > > can capture it all.
> >
> >MW: I find it is much more about not preventing it being captured.
> 
> The nice thing about a good ontology is that you can always add new 
> types when the need for them is recognized.    (016)

MW: Yes, that is why we took such care to make sure ours was extensible
(within the 4D paradigm).    (017)


_________________________________________________________________
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    (018)
<Prev in Thread] Current Thread [Next in Thread>