RE: true/false in RDF?

> -----Original Message-----
> From: Joshua Allen [mailto:joshuaa@microsoft.com]
> Sent: Wednesday, March 16, 2005 6:50 PM
> To: Geoff Chappell; Danny Ayers
> Cc: Seth Russell; semantic-web@w3.org
> Subject: RE: true/false in RDF?
> 
> > and into the data, you have to try to model your data in ways that
> will
> > allow the machines to perform meaningful reasoning. Given the current
> 
> > the only way to license inferences. Hopefully more expressive
> mechanisms
> > (e.g. rules) will join the party at some point so we don't have to
> 
> I can agree that this is the tradeoff.  I believe that taking dependency
> on RDFS, inference, and OWL today means that you chop off about 90% of
> your potential integration partners.  

Is there a necessary dependency? If you don't have a reasoner built in to
your app, it just means you have to realize more of the semantics in
application code. Either way, your app can read, write, and perform actions
based on the data. But suppose you define an ontology that captures as much
of the semantics of your data as possible (even if it's not used in your
app). Other non-reasoning app's seeking to interoperate with your data can
use the ontology as formal a description of what they need to implement in
code. Reasoning apps, OTOH, will be able to perform meaning-preserving
transformations and queries on your app data without any programmer
intervention.

Or do you think that it's the knowledge overhead/learning curve of rdfs/owl
that limit its use to a subset of the developer population?

[...]

> And I hate to see us making data modeling decisions based on
> technologies which are not mature and are at best enterprise-only;
> because for example, would we make the same modeling decision if rules
> were available today instead of OWL?  

When all you have is a hammer... i.e. when all of your reasoning capability
is class based, you're forced to try to squeeze everything in to that model.
Some things fit perfectly, some not so well, and some not at all. So sure, a
more expressive language would allow you to say things that you can't say
today and might enable to you to say some things in a more natural manner.
But isn't one of the points of having clearly defined semantics that we'll
be able to represent anything we model now in a new more expressive world
(and that translation to the new representation should be completely
automatable)?

-Geoff

Received on Thursday, 17 March 2005 15:00:45 UTC