Re: Ontologies for describing Enterprise Architectures
I agree with John's comments below on the importance of first 'defining'
certain key terms. For clarification, I assume a 'definition' is a set of
axioms plus a gloss and perhaps some examples rather than just the gloss
itself.
One can't state too strongly that 'function' (outside of mathematics and
certain programming languages) is a mess. I would add that 'objective',
'purpose', and 'goal' are not only basic and difficult, but often used and
confused. I have some notes somewhere about terms that have been
particularly troubling to me when I've tried (so far with little concrete
output) to axiomatize certain management process standards I worked on in
the past.
Enough said. I'd like to see John's program to pick a few key terms and
define them carried out. I'd be happy to participate. (In my naivete when
I first became interested in this group I thought that something like this
would be the main goal of a SUO. I was shocked when I read some of the
early discussion concerning how many thousands of terms would be needed.)
Best,
John Velman
On Wed, Nov 10, 2004 at 05:54:23PM -0500, John F. Sowa wrote:
> Jim,
>
> I certainly agree with that point:
>
> > I would agree that IEEE 1471 is the standard for
> > enterprise architecture descriptions. However, I
> > would question its relevance for describing Enterprise
> > Architecture Ontologies. IMHO, IEE 1471, FEA, TOGAF and
> > C4ISR are frameworks. The problem with a framework is
> > that it does not provide explicit set of guidelines
> > for creating definitions.
>
> That is another good point:
>
> > It has been my experience that EA projects rise and
> > fall on the ability to reach consensus on definitions
> > (e.g. objective, function, IS-A relationships,
> > IS-Part-Of).
>
> This leads to a suggestion of how the SUO should
> proceed:
>
> 1. Before trying to reach agreement on an official
> standard upper ontology, it would be much simpler
> to standardize the definitions of certain terms
> that are necessary to define any ontology at
> any level.
>
> 2. Such a standardization would be a prerequisite for
> any upper ontology, and it would be essential for
> making it possible to merge any theories in a family
> of ontologies or to select among possible alternatives.
>
> Is-an-instance-of, Is-a-kind-of, and Is-a-subtype-of,
> are three essential relations that I believe we can
> reach consensus on rather quickly.
>
> Is-a-part-of raises some serious questions because
> there are many different axioms for mereology.
> One solution is to take the most general definition
> (i.e., the one with the smallest number of axioms)
> as the primary, and to define a small hierarchy
> of others that depend on which additional axioms
> are accepted for each one.
>
> But I'm not sure what you mean by "objective"
> and "function". I understand the term "function"
> in its mathematical sense, but people frequently
> use the term in talking about the "function" of
> a refrigerator. Trying to define "function" in
> that sense is asking for quagmire duty.
>
> And talking about the "objectives" of an enterprise
> or some department of an enterprise is very important,
> but very difficult to define. Just trying to define
> "enterprise" is nontrivial. In fact, you probably
> cannot define the notion of "enterprise" without
> getting a good handle on the notion of "objective".
>
> John
>