Richard,
As you note, one may take different views of individual
entities different times, or even have multiple ways of classifying something
at the same time. But no matter how many different views one may want to
take, the type hierarchy should remain the same. In your example below,
it was not clear whether the terms “car” “chassis”,
“engine” were being used as instances of such things, or were
actually types in the hierarchy. If the latter, modifying the hierarchy
to adapt to different views is likely to lead to utter chaos.
The better tactic is to include all variant views
in the same hierarchy, and, when communicating a representation, indicate which
view is being used. Unless the views are logically incompatible, they
should be translatable into each other, so that either view may be used by any
application.
When discussing “parts”, if one
wants the machine to do automated inferencing, it is important to decide
whether a “part” is essential, in which case the whole does not
exist without the part, and if essential, whether it can be removed from its
functioning location in the whole and still be considered a “part”
of that whole. It is perfectly reasonable to disassemble a car in a
garage, and still say that “the car is in the garage”. But
that is a convention that one must decide on when creating an ontology and an
application that uses it.
In order to represent such nuances, I find that
multiple “part” subrelations are useful, each meaning something
different.
A general issue in creating relations is to
decide whether it is an instance-level relation or a type-level relation.
In the latter case, one might want to say. e.g. that every living human
has a brain – which is different from saying that some specific human has
a brain. Being very explicit is necessary when the ontology is to
be used for automated reasoning.
Pat
Patrick Cassidy
MICRA, Inc.
908-561-3416
cell: 908-565-4053
cassidy@xxxxxxxxx
From:
ontolog-forum-bounces@xxxxxxxxxxxxxxxx
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Richard H.
McCullough
Sent: Saturday, September 06, 2008 10:40 AM
To: [ontolog-forum]
Subject: Re: [ontolog-forum] Fw: Thing and Class
I have
one remaining question -- what does your
concept
hierarchy look like?
In my
tabula rasa hierarchy, parts are characteristics
of the
entity that they are a part of. So you might
But when
the car is disassembled, it would look like this
Thus my
concept hierarchy - my point of view - changes
with
time, as the parts are assembled or disassembled.
In your
case, I would think the designers, manufacturers,
sales
personnel, etc. would have different hierarchies.
Dick
McCullough
Ayn Rand do speak od mKR done;
mKE do enhance od Real Intelligence done;
knowledge := man do identify od existent done;
knowledge haspart proposition list;
http://mKRmKE.org/
-----
Original Message -----
Sent:
Saturday, September 06, 2008 3:44 AM
Subject:
[ontolog-forum] Fw: Thing and Class
>
> Dick,
>
> In engineering practice (at least as described through ISO
> 10303), a part is either an inseparable component or an assembly. An
> assembly consists of sub-assemblies (which are assemblies) and
> components. The definition of assembly is recursive.
>
> A product is a part. In ISO 10303, a part is a subtype of a
> product, and this is quite distinct from a product concept. Thus, a
> product concept might be "Range Rover", a four-wheel drive
vehicle,
> which is delivered as physically as a part RR1-801/1, or RR1-801/2,
or
> RR1-803/1, or RR1-805/1, or RR2-810/1, etc, where in RRx-yyy/z, the RRx
> is the generic part number, the yyy is the manufacturing assembly suffix
> and the /z is the version number. In theory, all parts with the number
> RR1-801/1 are completely interchangeable, and should be interchangeable
> with parts R1-801/2 (which differs from R1-801/1 in matters not
> affecting form, fit or function), however will have some form, fit or
> function difference from RR1-803/1.
>
> The occurrences of the Range Rover concept will also be stamped
> with a serial number. A typical maintenance manual will inform you of
> changes made in the product standard in terms of the serial numbers that
> the change effects - e.g. from 1-100,000 it has a 2.3 litre engine, and
> from 100,001 onwards the vehicle has a 2.5 litre engine. In practice,
> minor changes which internally generate new part numbers will not be
> exposed at the level of the maintence manual.
>
> Of course, the real situation is rather more complicated, but in
> matters of detail, rather than of substance (in the context of this
> discussion).
>
> The point being, that a change approval creates a commitment to
> create a new part (car, wheel, bolt, etc) which is either a revision of
> an existing part (form, fit, function interchangeable) or a new part. In
> the case of an assembly, this may be because it incorporates a new part,
> or a new combination of existing parts. Conversely, creating a new
> component part has no effect on end product unless it is embodied in the
> chain of assemblies which contributes to an actual part (end product),
> which in turn requires that a new version of these assemblies is
> produced.
>
> The issue "is a representation of" versus "is-a" versus
> "is-an-instance-of" is a question of how we ground these
relationships
> and how we ground the concept of part. I ground the concept
"part" as a
> "form, fit, and function equivalence relation", such that two
part
> designs (which specify the part) are equivalent if the physical part
> they specify is form, fit and function interchangeable. This means that
> the physical parts are an instance of the class "part", and that
a
> design is a specification for the part. That is, the meaning of the
> concept "part" is grounded in the results of using individual
parts.
> Consequently, a design office designs many parts, some of which are
> instantiated by the manufacturing department as physical parts. In the
> design office, parts instantiate the design process outputs, and are
> treated as separate individuals, while in the manufacturing world, a
> part is a class for producing physical parts.
>
> I hope this answers the question.
>
>
> Sean Barker
> Bristol, UK
>
> -----Original Message-----
> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
>
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Richard H.
> McCullough
> Sent: 05 September 2008 11:24
> To: [ontolog-forum]
> Subject: Re: [ontolog-forum] Thing and Class
>
> Sean & Dan
>
> It is hard for me to understand what this discussion is about.
> When I see things like
> part is described by design
> and
> part is instance of design
> I wonder if you are missing the whole concept of
> part is part of design
> i.e., the part-whole relation.
>
> In Sean's last email, time dependence is mentioned, and I wonder -- are
> you now talking about a part-whole relation which is time-varying?
>
> Dick McCullough
> Ayn Rand do speak od mKR done;
> mKE do enhance od Real Intelligence done; knowledge := man do identify
> od existent done; knowledge haspart proposition list; http://mKRmKE.org/
>
> ----- Original Message -----
> From: "Sean Barker" <sean.barker@xxxxxxxxxxxxx>
> To: <ontolog-forum@xxxxxxxxxxxxxxxx>
> Sent: Friday, September 05, 2008 12:25 AM
> Subject: [ontolog-forum] Fw: Fw: Thing and Class
>
>
>>
>>
>> Dan,
>>
>> While I would agree that, say, the CAD model for a part
>> describes the shape of a part, the issue is not one of design but of
>> configuration management. In particular, the criterion for being a
> part
>> A123 is that it is fit, form and function identical to the
"typical
>> part" A123. The design is an "ontological commitment"
that some class
> of
>> thing exists (will exist). To reject this is to reject the concept of
>> "is-a" and of labelling things with the concepts they
instantiate.
>>
>> Conversely, penguins do not stop being penguins just because some has
>> sequenced their DNA (written down their design).
>>
>> The fact that engineering systems are concerned with coming-to-be and
>> ceasing-to-be suggests that
>> engineering ontologies must use a temporal logic. In fact, many
>> engineering
>> systems
>> are based on effectivities and change notices. The first explicitly
>> identifies what components
>> make up a product at a particular time or at a point in the product
> run,
>> while th second
>> controls when the definitions are changed.
>>
>> Sean Barker
>> BAE SYSTEMS - Advanced Technology CentreBristol, UK
>>
>> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx
>>
[mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx] On Behalf Of Dan
> Corwin
>> Sent: 01 September 2008 19:46
>> To: [ontolog-forum]
>> Subject: Re: [ontolog-forum] Fw: Thing and Class
>>
>>
>>
*** WARNING ***
>>
>> This mail has originated outside your organization, either from an
>> external partner or the Global Internet.
>> Keep this in mind if you answer this message.
>>
>> No magic here, just typical abstract and concrete objects.
>>
>> Sean Barker wrote:
>>
>>> 2) The product of a design office is designs, instances of the
> general
>>
>>> mathom "design". In the DO, any class/type structure
applied to a set
>>> of designs is a generalization of the set of design instances -
>>> designs are
>> > not classes for anything.
>>
>> A "design" is surely an object in the world of information.
>> It describes something, which you portray below as concrete.
>>
>>> The product of a manufacturing organization is parts, each of
>> > which is an instance of a design.
>>
>> Wrong. Each "part" may be based on the
"design", but their
>> relation is described/describes, not instance/class.
>>
>> regards,
>> Dan Corwin
>>
>
>
>
>
> _________________________________________________________________
> 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
>
>
>
|
_________________________________________________________________
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 (01)
|