[wg11-owl] Goal of the review

[Ed Barkmeyer]

ewallace at cme.nist.gov wrote:

> I doubt that SC4:1992 is comparable to the current state of OWL tool support.
> This is because OWL leveraged past work in RDF and Description Logics, and
> because development was fueled by major research initiatives on both sides of 
> the Atlantic.  There are a number of academic and commercial implementations
> of OWL reasoners and triple store development frameworks.  The quality of
> some of the these tools is quite good and is having a strong influence on
> the functionality and APIs of their competitors.  If and how this will
> translate to manufacturing application support for OWL is anyones guess.

To be careful, there were in 1992-3 half a dozen academic and 
quasi-academic EXPRESS compilers and Part 21 toolkits, and one or two 
fledgeling commercial efforts.  Yes, you can find tools to parse OWL/XML 
and make triple-stores and feed various GUIs for "ontology development", 
which is all about making and capturing models.  But the result is still 
model databases.  I agree that the quality of OWL tooling is better than 
the quality of EXPRESS tooling in 1992, because the quality of GUI and 
database and programming tooling is generally better.

Yes, there are reasoners, almost all of which existed before OWL, but 
there were databases and internal product models, and tools to do 
geometric reasoning before EXPRESS and AP203, too. The question is: 
What problem do we want to solve that using OWL and these tools can solve?

You will pardon my view that we hardly need yet another way to write 
down taxonomy graphs and information models.  The fact that a whole 
collection of XML freaks and AI academics are learning to do information 
modeling with a different set of terms and tools doesn't impress me. 
Yes, OWL has capabilities we are not used to, but it also lacks 
capabilities we are used to.  The question is:  What is it good *for*?

This study is yet another "state-of-the-art solution" looking for a 
problem.  So I raised the two questions I think we most need to answer: 
  Do we believe that the OWL hype IS the problem we need to solve?  Or 
do we really have problems we think a "reasoning tool" would be able to 
solve, so we can do a useful technical evaluation?

The problem proposed by David Leal is:  Given an APxxx or PLIB "product 
model" rendered into OWL, and a set of product instances documented 
according to that model, and a set of requirements for a part, to 
determine whether any "product" in the information base meets those 
requirements.  So: to what extent can OWL be used to capture models and 
requirements for form, fit, function?  Can some OWL reasoners find 
matches from models of form, fit and function of individual parts 
(involving values for classifications and measurements and features)? 
Can they do that somehow better than SQL queries (tests for equal, tests 
for values within range, tests for values in sets)?  That is the kind of 
thing I would have expected.

-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."