[Top] [All Lists]

Re: [ontolog-forum] Ontology and methodology

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: "John F. Sowa" <sowa@xxxxxxxxxxx>
Date: Mon, 19 Mar 2007 17:39:56 -0500
Message-id: <45FF113C.4050006@xxxxxxxxxxx>
Dear Matthew, David, Chris, Pat, John,    (01)

MW> So Matthew West the employee of Shell is only the same
 > thing as Matthew West the person if Matthew West is an
 > employee of Shell for the whole of his life.    (02)

Nonetheless, it is always true in both 3-D and 4-D models
that every instant of time when MW is an employee, MW is
also a person.  Therefore, Employee < Person.    (03)

MW> The employee instance is not a person-for-the-whole-of-their-life,    (04)

Of course.  But every employee instance is a part of the person
instance.  That is exactly what is implied by Employee < Person.    (05)

DD> Simply said, identifying whether something is a 'natural kind'
 > or a role might be very hard in some cases and completely depends
 > on the level of granularity of your ontology or the background
 > of the modeler.    (06)

This is one of many reasons why trying to draw a hard-and-fast
distinction that holds for every possible case is not easy.
Therefore, it makes more sense to put all the concept types into
a single hierarchy and leave it underspecified, if necessary.    (07)

CP> Your notion of employee does not distinguish between employers
 > - and it is also temporalised - one is an employee at a time.    (08)

No.  The statement X<Y is about *types*, not about *individuals*,
and the *definition* of a type is independent of space and time.
What is time dependent is whether a particular individual is
an instance of a particular type at a given time.    (09)

CP> It seems that implicitly mean that *at time t1*, X < Y means
 > that every instance of X is an instance of Y. Maybe better to
 > say that X <t1 Y means that at t1 every instance of X is
 > an instance of Y.    (010)

No. To say Employee<Person means that at *every* point in time
and place, if x is an employee, then x is a person.  It doesn't
matter whether x is an employee of more than one company at
that time, the simple statement "x is a person" is still true.    (011)

CP> I am not sure how this would deal with employees with
 > multiple employers at the same time. It looks as if it
 > would have problems. Could you explain.    (012)

Please remember that Employee < Role.  Therefore, one cannot
be an employee without being in a role.   If you are in the
same type of role with two different companies, then you
are playing two roles at the same time.  You have not become
two different individuals.    (013)

CP> However, this is not the notion I find when I look at
 > systems.  Typically a system will say who the person is
 > employed by, when he/she started employment, and if
 > appropriate when they left employment.    (014)

Of course.  And during the entire time, that person did
not cease being a person.    (015)

CP> a bishop would be a sub-type of person.  If James were
 > a bishop, you would say that there is only one James who
 > is both a person and a bishop (at time t1).    (016)

Of course.    (017)

CP> If one left ones property to the Bishop, then when James
 > dies in office, the property would remain the property of
 > the bishop and not pass to James's heirs.    (018)

That depends on how you stated the deed, either explicitly
or implicitly.  There are usually legalistic phrases attached,
such as "ex officio", to indicate how the money is given.    (019)

If a clerk at McDonald's gives you a hamburger and asks for
payment, the money is implicitly designated for McDonald's,
not the clerk.  But if the clerk happens to be your niece, you
can give her a gift for her birthday while you're buying the
burger, and it won't go into the money drawer.  But there is
still only one person there -- not a separate clerk and niece.    (020)

PH> Seems to me pretty obvious that I am a person and I am also
 > an employee, and also that there is just one of me.    (021)

I'm glad that we're on the same side in this debate.    (022)

JFS>> If I'm assigned a task as an employee, it's my task, not
 >> the task of some virtual employee.    (023)

JB> Evidently not as settled after all: this is the qua-individual
 > debate, which is alive and well last I heard.    (024)

But note the "qua" -- that is a sign that we're not talking about
the number of individuals, but about the number of different
roles that a single individual may be playing at the same time.    (025)

JB> The usual counter-discussions to the above point are the legal
 > ones I think: who is responsible and who can get sued for what?
 > Or the passengers-on-Airline-X examples: how do you count them?
 > And what have you counted once you have counted them? Seems like
 > there are lots of virtual artefacts waiting to take their place
 > in the discussion...    (026)

But those "artefacts" are multiplicities of roles, not of
individuals.  There is no problem about counting the number of
people on the plane.  But when a plane has multiple flight numbers,
it can be hard to determine which passenger-role any particular
individual is playing at that moment.    (027)

JFS>> Subtype has a very clear and simple definition:  X < Y
 >> means that every instance of X is an instance of Y.    (028)

PH> I would prefer to say that it implies this, but subtype
 > is actually somewhat stronger.    (029)

Sorry, I should have said "implies" rather than "means".  It's
a one-way implication, not an "iff".    (030)

JFS>> It implies that the entire spatiotemporal extent of Employee
 >> is included in the spatiotemporal extent of Person.    (031)

PH> Not unless you conflate class membership with temporal parthood.
 > I think Matthew is keeping these ideas separate.    (032)

This may be the crux of the distinction.  Pat and Matthew are equating
types with classes (which I would interpret as extensional set-like
things).  But I interpret types as intensional, name-like things.
Therefore, I interpret X<Y as implying that for every space-time point
where the type label X is appropriate, the type label Y is appropriate.    (033)

FP>> In one of the extensions of DOLCE, a role is
 >> considered a non-physical social object.    (034)

PH> I think that is a rather special interpretation of 'role'.
 > It had better be, as the idea of 'social object' makes no sense
 > at all in many domains of application.    (035)

As an example, consider the definition of "planet" vs. "satellite".
There are some satellites in the solar system that are larger than
the planet Mercury.  To use my criterion, you can't tell whether
they are planets or satellites just by looking at them in isolation.
Therefore, the distinction must be by their role in relation to
other bodies in the solar system.  But there is an objective test
for making the distinction.    (036)

PH> Everyday common sense is simply one more area of expert
 > specialization, like protein folding or cell biology: we don't
 > think of it that way only because all competent adults are
 > experts, like chess grandmasters in this particular domain.
 > But it is just as peculiar and idiosyncratic as any other,
 > and ontologies designed for it don't automatically transfer
 > to other domains.    (037)

I agree.  One generation's cutting-edge research is the next
generation's common sense -- for example, computer technology.    (038)

FP>> If a certain entity which was previously identified as
 >> chair, is never used again for sitting but only for, say,
 >> a step for changing light bulbs, it stops playing the role
 >> of being a chair and is hence only playing the role of
 >> a step for changing light bulbs (?).    (039)

PH> It would still be a chair, however (or at least one could
 > reasonably claim this.) We would refer to it that way, and
 > it would be listed that way in a product catalog.    (040)

Yes, there is a characteristic shape for a chair.  Fragile
antiques in a museum are chairs even though nobody is allowed
to sit on them.    (041)

PH> I actually prefer to have a clear-cut abstract/spatiotemporal
 > distinction and firmly relegate all mathematically described
 > collections to the former category. Its just easier to remember,
 > basically.    (042)

There's a lot to be said for definitions that are easy to remember
and minimize confusion.  Whitehead used the term "eternal entity"
for anything that does not have spatiotemporal coordinates, but
it led people to think he was talking about heaven and angels.
I'm happy with the abstract/spatiotemporal distinction, and I
would put all mathematical structures into the abstract category.    (043)

John    (044)

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx    (045)

<Prev in Thread] Current Thread [Next in Thread>