Matt, (01)
This is an excellent discussion as it hits upon some key issues that
don't involve interminable squabbles about metaphysics (not that there's
anything wrong with that :-D) and are central to the determination of
ROI for the use of upper ontology. See below. (02)
West, Matthew R SIPC-DFD/321 wrote: (03)
>>LEO: There may be those who have
>>promoted wider visions of those models, but I expect 99% of database
>>folks (theorists and practitioners) will say: only obliquely are we
>>concerned with the "real world".
>
> MW: I've been developing data models for 20+ years now and even in the
> early days the numbers were much better than that where I come from.
> Nowadays it is hard to find someone developing data models (let me be
> explicit here, logical or conceptual data models and not so much database
> schemas) who does not see themselves as being about concerned about
> the "real world". Apparently we don't inhabit the same "real world". (04)
I am in full agreement with Matt on this point. I spent only 10 years
on data modeling but what he says was also my experience. In each
project I was involved in, the parties involved were trying to build
models of real world (no scare quotes) entities. They were just trying
to do it with OO/ER. (05)
> MW: Another key point is the need for an ontology to be expressed in
> a logic. Clearly this is necessary if your purpose involves inferencing
> over the ontology, but frankly this is a minority sport. There is vastly
> more ontology embedded in database schemas, their contents, and the
> procedural code that operates on them than in ontologies expressed in
> terms of first order logic. (06)
I think this is a terminological clash, Matt. Leo is explicitly talking
about "ontologies" -- documents written in some formalism. You seem to
be talking about "ontology", the philosophical discipline. And yeah,
from what I've seen, there has been some good ontology done in the data
modeling world, albeit with crippled tools (procedural code is not a
good place to house your reusable semantics). (07)
> MW: Now much of this is "crappy" ontology, but so is much of what is
> expressed in FOL. (08)
Precisely. There has come to be an elan associated with the use of
logic because of the relative smaller priesthood that understands it vs
more traditional approaches. But its just a tool and lots of bad things
get built with good tools. This, BTW, is a MAJOR engineering reason to
adopt a ULO approach -- by constraining the "imaginations" of domain
ontologists, fewer mistakes get made. This has been our experience in
every project we've done. (09)
> The result, however, is that the benefits to be gained
> in applying better ontolgy in business applications in the relatively
> short term is much greater than in the relatively green fields of
> inferencing, and this is not dependent on being expressed in a logical
> form. (010)
Also an excellent point. There's a lot of loose talk about inferencing
that crops up in discussions about ontology or ontologies. Seems to me
that ontology, in this sense, is often confused with expert systems --
we're expecting an "ontology" somehow to be smart. Better to persue the
more modest goal of using ontologies to organize data. (011)
Of course the road from ontology to business application for organizing
data need not be such a long one. To say more would be an advertisement
for Ontology Works :-D (012)
-- (013)
Bill Andersen
Chief Scientist
Ontology Works, Inc. (http://www.ontologyworks.com)
3600 O'Donnell Street, Suite 600
Baltimore, MD 21224
Office: 410-675-1201
Cell: 443-858-6444
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/uos-convene/
To Post: mailto:uos-convene@xxxxxxxxxxxxxxxx
Community Portal: http://ontolog.cim3.net/
Shared Files: http://ontolog.cim3.net/file/work/UpperOntologySummit/uos-convene/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?UpperOntologySummit (014)
|