[wg11-owl] Re: Paper: Criticism of the STEP Application Module approach

[Ed Barkmeyer]

David Leal wrote:

> The paper is attached.

Thanks for that.

> Unfortunately I do believe that - the negative assessment is true in many
> cases. The STEP methodology can fix-up ill-defined concepts by specific
> usages within an AP, but as soon as the concepts are converted into an
> ontology, the failings are exposed and need to be fixed.

Exactly.  Moreover, we have then the opportunity to determine whether 
the "interpretation" in the AP, which should be a consistent 
micro-theory, is in fact consistent with the "fixed" general interpretation.

> The creation of an
> 'ontology' of ill-defined concepts is a complete waste of time.

I agree fully.  But we need to be careful to distinguish "well-defined 
very general concepts", like much of the SUMO stuff, from "ill-defined 
concepts".  It is my perception that SC4 was trying to do the former, 
but, as a body, lacked the linguistic/semantic discipline to do that 
well (as evidenced by bad definitions and some misguided half-educated 
dicta from the "Quality committee").  I have no objection to converting 
bad definitions to good ones, so long as the intent of those definitions 
does not change.  I do object to converting intentionally general 
definitions to narrower ones, even when the latter is "what we should 
have meant".

Further, in making ontologies from SC4 models, there is nothing sacred 
about IRM concepts.  IRM concepts, AIC concepts, module concepts and AP 
concepts are all Given.  The object is to fashion consistent ontologies 
from the set of Givens.  In the long run, what counts is that we have 
good ontologies for the AP, AIC, and module models.  If consistency in 
that universe can be attained by using the apparent ontology for the IRM 
concepts, well and good.  But if it can't, valueless IRM classes and 
properties can be discarded in favor of valuable (and recurring) AP, 
AIC, and module interpretations.

The argument about the term "product", for example, is that it is 
intentionally broader than was necessary, and that it merges (the "in" 
word is "confabulates") two importantly different concepts - a thing and 
the design for the thing.  We can't "fix" either of those errors in Part 
41 without risking inconsistency with some AP interpretation.

But the "SC4 ontology" need not use "product" from Part 41 at all, if 
there are two inconsistent interpretations that are used in APs.  We 
state the two AP definitions, prove the inconsistency, and discard the 
Part 41 term, in favor of the two AP terms or two terms, each being 
"product" qualified by some word that distinguishes the AP definitions.

My concern is that the "SC4 ontology" match, not rework or redefine, the 
SC4 models.  It does not have to "match" the broken parts of the SC4 
architecture.

-Ed

-- 
Edward J. Barkmeyer                        Email: edbark at nist.gov
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8264                Tel: +1 301-975-3528
Gaithersburg, MD 20899-8264                FAX: +1 301-975-4694

"The opinions expressed above do not reflect consensus of NIST,
  and have not been reviewed by any Government authority."