SUO: Applications & ontology [was: Negotiation Instead of Legislation]
On 2/23/02 20:29, "Douglas McDavid" <mcdavid@us.ibm.com> wrote:
> So, let me pose the question to the list, especially David Whitten,
> Ian Niles, and Scott Farrar:
>
> What applications do you have in mind when you debate constructs such
> as:
>
> (subclass WrittenLinguisticExpression LinguisticExpression)
> (subclass SpokenLinguisticExpression LinguisticExpression)
> (subclass GesturalLinguisticExpression LinguisticExpression)
>
> How do you envision that these assertions will interface with other
> ontologies? What kinds of application-level ambiguities will be
> able to be disambiguated by this kind of structure? What are the
> practical purposes that will be served, in what kinds of working
> domains, for these particular constructs?
>
> These questions can apply to all content of the SUMO, or alternative.
Doug,
Your (and Leo's) point about the need to focus on applications is a very
important one. However, I think your use of the example above may be off
the mark. For example, you might ask which applications I have in mind when
I write (in SUMO-speak) some of seemingly equally abstract axioms of
Guarino and Welty that are part of their OntoClean methodology:
(subclass NonRigidProperty SortalProperty)
(subclass RigidProperty SortalProperty)
(subclass AntiRigidProperty NonRigidProperty)
(disjoint RigidProperty NonRigidProperty)
...
[Pat - No squawking about sortals! It's just an example!!!]
These have proven quite useful, NOT in building applications, but in
building ontologies that in turn form the basis of good an maintainable
applications. My company's tools generate databases, programming
interfaces, and other software objects from these ontologies. As an
application - our tools exploit such properties directly (unfortunately how
we do that is proprietary), so at least *one* application benefits from
these kinds of things...
All I can say is that the quality of the resulting ontologies is higher than
that obtained by introspective hacking, especially when the models become
large and manual checking of constraints becomes impossible. Actually, to
back this up, I have an existence proof of an episode where, if I had
applied the above axioms to a RDB schema design, I could have saved the US
taxpayers about $500,000 on a project I once worked on. Since some of you
have heard that story before and it's kind of detailed, I won't repeat it
here online, but will do so on request from anyone interested.
Cheers,
.bill
--
Bill Andersen
Chief Scientist, Ontology Works
1132 Annapolis Road, Suite 104
Odenton, Maryland, 21113
Mobile: 443-858-6444
Office: 410-674-7600
Web: http://www.ontologyworks.com