RE: [SE] Suggestion of new note

Mike

Just a remark about your use of "sets" in the second slide

"OWL: classes are sets of individuals"

If one is to understand here "set" as in "set theory", this looks plainly wrong to me and
could be at least misleading to people familiar with set theory.

A set has elements, a class has instances. Both can be defined in intension or extension.
But whatever its definition, extensional or intensional, the extension of a set (finite or
not) is determined once for all. In some cases there might be no known easy way to compute
the extension. Take the set of all prime integers. The extension of this set is very
difficult to determine and compute, as everybody knows. But it is nevertheless once for
all completely *determined* by its intensional definition. And it's the same with every
set you properly define in set theory.

At the opposite the extension of a class is not necessarily completely determined by its
intensional definition. Take "foaf:Person", no one can reasonably claim this defines a
set, since its extension is not bound by its intentional definition. You will forever be
able to add new instances with every newborn baby, not to mention the undiscovered dead
and imaginary instances. Unless you claim that all possible instances of "Person" are
already defined somewhere in the metaphysical blue, and we'll only discover them along the
way like prime numbers, but I won't follow you on this track :))

Beyond vocabulary, it's unclear to me what kind of opposition you want to capture in
defining OO classes as types vs OWL classes as sets. Could you expand on this point?

Regards

Bernard

----------------------------------
Bernard Vatant
Mondeca Knowledge Engineering
bernard.vatant@mondeca.com
(+33) 0871 488 459

http://www.mondeca.com
http://universimmedia.blogspot.com
----------------------------------

> -----Message d'origine-----
> De : public-swbp-wg-request@w3.org
> [mailto:public-swbp-wg-request@w3.org]De la part de Uschold, Michael F
> Envoye : mardi 4 octobre 2005 21:37
> A : Holger Knublauch
> Cc : public-swbp-wg@w3.org
> Objet : RE: [SE] Suggestion of new note
>
>
> Here are my latest slides on the difference between object-oriented and
> OWL.
>
> Mike
>
>
> ============================================
> Mike Uschold
> Tel: 425 865-3605              Fax: 425 865-2965
> ============================================
>
>
>
> >  -----Original Message-----
> >  From: Holger Knublauch [mailto:holgi@stanford.edu]
> >  Sent: Tuesday, October 04, 2005 10:13 AM
> >  To: public-swbp-wg@w3.org
> >  Subject: Re: [SE] Suggestion of new note
> >
> >
> >
> >  Hi Mike,
> >
> >  thanks again for your comments.  Sorry I could not respond
> >  earlier - I
> >  am currently in the (time consuming) process of looking for
> >  a new job :)
> >
> >  Comments below.
> >  Holger
> >
> >
> >  Uschold, Michael F wrote:
> >  > Holger,
> >  >
> >  > 1. This document is not listed as a deliverable in the TF
> >  Web page, is
> >  > that intentional?
> >
> >  I will send the current draft to Jeff soon - it should be listed.
> >
> >
> >  > 2. A few more quick thoughts on the table comparing
> >  OBJECT-ORIENTED
> >  > and OWL.
> >  >
> >  > This is wordy and hard to follow:
> >  >
> >  > Instances can only take values for the properties attached
> >  to its type.
> >  > Values must be of the correct types defined for the properties.
> >  >
> >  > Any instance can take arbitrary values for any property,
> >  but this may
> >  > affect what reasoners can infer about their types.
> >
> >  I have cleaned this up, following your suggestion in the
> >  follow up email.
> >
> >
> >  > This suggests that OWL is at a disadvantage, it can't do
> >  privacy. You
> >  > need to emphazize the OWL Advantage that it makes it
> >  possible to link
> >  > ontologies from all over the place, and privacy can
> >  probably be added,
> >  > so is not a fundamental difference.
> >  >
> >  > Classes can encapsulate their members to private access.
> >  >
> >  > All parts of an OWL/RDF file are public and can be linked to from
> >  > anywhere else.
> >
> >  I wouldn't read the current statement as a negative statement.  I
> >  mention that all parts "can be linked to", which sounds like an
> >  additional feature to me.  However, I don't really see how
> >  privacy can
> >  be added in OWL.  Neither is better or worse, but both approaches
> >  fulfill their design goals.
>
> I'm not sure we're in agreement, but it is not a big point. Might be
> easily addressed by slight rewording to make it less likely for someone
> to give it a negative interpretation.
>
>
>
> >  > Also, the long list is hard to make sense of, there are nice
> >  > categories that would be good to use to organize the
> >  items. Even if
> >  > there is just one entry in the category, it highlihts the
> >  topic making
> >  > it easier to
> >  > understand:
> >  > * Classes and Instances/Individuals
> >  > * Properties, Attributes and Values
> >  > * Errors and Consistency checking
> >  > * Maturity
> >  > * Worldliness (open vs. closed)        [not serious about
> >  the category
> >  > name :-)
> >
> >  This is a great idea and I have done some partitioning for
> >  the next draft.
> >
> >
> >
>

Received on Thursday, 6 October 2005 10:21:26 UTC