Bill Burkett wrote:
> Something that you don't mention, however, is that the value often touted for
>ontologies and the use OWL is that because it's logic based you can do
>reasoning/inferencing with OWL representations and discover "new" information.
> Unfortunately, to get the semantic expressiveness that you see in semantic
>data modeling languages, you need to use OWL-Full - and there is no guarantee
>that you successfully reason/inference over OWL-Full representations (or at
>least bring that reasoning/inferencing to completion over a finite amount of
>time.)
> (01)
This is true of _some_ of the expressive capabilities of SDM languages.
If you are thinking of EXPRESS and OCL as "semantic data modeling"
languages, then yes, you can write very complex integrity rules in those
languages. But EXPRESS, for example, forces the use of complex
expressions to express a number of simple relationships for which OWL
has explicit syntax. And OWL can directly capture all of the common
semantic relationships (as distinct from all kinds of integrity
constraints). (02)
> Thus, the main value of OWL is compromised and it becomes essentially
>equivalent to other semantic data modeling languages.
>
> Am I missing something, or are the other values that can be derived from an
>OWL representation if you set aside the advantage of being able to
>successfully reason over the it?
> (03)
The main advantage is that the semantics of the OWL constructs
themselves is formally defined. We know exactly what a particular OWL
expression means; we don't have to talk about modeling styles and the
conventions used by group X. This is a major difference from UML and
EXPRESS and IDEF1-X, in which one can never be certain whether a given
model element represents a concept or a data structure. Both NIAM/ORM
and SDM had formal semantic models defined for them "after the fact",
which unfortunately doesn't mean that models made in those languages had
exactly those interpretations. In the case of OWL, that formal
semantics is part of the definition of the language -- if you use an OWL
construct with another intent, your model does not mean what you think
it means. (04)
Now, I think that "setting aside the ability to reason" is rather like
ordering "a hamburger -- hold the beef". For the kind of thing Len is
doing, for example, you don't need OWL/Full in most cases. Much of that
semantic analysis is about classification and the relationships between
classifications of things and data structures representing the things.
NIST and other experiments in "semantic mediation" indicate that OWL/DL
(and "extended tableaux reasoning") is fairly powerful in this regard.
In most cases, perhaps 5% of the knowledge content associated with data
schemas cannot be captured in OWL/DL. But the impact of that 5% on
conversions or analysis is much harder to judge. So if Bill needs 100%
knowledge transfer, then OWL/DL and tableaux reasoning will fail his
test; but if you are trying to get useful practical results, OWL/DL can
go a long way. (05)
(Now, since I know Bill from the ISO TC184 EXPRESS data modeling
activities of the last 20 years, I realize that many of his "semantic
data models" are probably EXPRESS models with ornate rulesets. Many of
those are built around geometry, which one would never want to do in
OWL. So I freely admit that OWL/DL won't go a long way in capturing those.) (06)
> P.S. I'd be very interested in seeing the presentation you mention - is it
>something you can distribute? (at least to me? :-)
> (07)
FWIW, I did the presentation on one of the Ontology Forum Webinars in
2007, and the slide set (and the recording, Peter being so well
organized) is on the wiki at:
http://ontolog.cim3.net/cgi-bin/wiki.pl?ConferenceCall_2007_04_12 (08)
-Ed (09)
--
Edward J. Barkmeyer Email: edbark@xxxxxxxx
National Institute of Standards & Technology
Manufacturing Systems Integration Division
100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528
Gaithersburg, MD 20899-8263 FAX: +1 301-975-4694 (010)
"The opinions expressed above do not reflect consensus of NIST,
and have not been reviewed by any Government authority." (011)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: 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 join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (012)
|