Re: how to proceed with building OWL [was Re: semantics: layering ...]

At 8:00 AM -0500 3/1/02, Peter F. Patel-Schneider wrote:
>	Constructing Web Ontology Languages
>				or
>	Why Jumping on Four Horses is a Good Recipe for Quartering



Peter- I think I agree with everything you said in your message, 
except the overall implication that somehow layering drives the whole 
process.  It is a necessary player, but it seems to me the language 
functionality, driven by the work we've been doing in use cases and 
reqs could also be put up as the main driver.  here's my argument


1) Let's assume the group decided that RDFS was the perfect ontology 
language and needed no new language features.  In this case, layering 
with RDFS is perfect and the problem goes away

2) Let's assume the group decided the entire DAML+OIL functionality, 
plus more is what we want.  In this case, the layering issue, and how 
we deal with it, becomes important.

So somewhere in the middle between no new features and lots of new 
features will be language designs that are perfectly compatible (i.e. 
no paradoxes), that are not as compatible (i.e. paradoxes abound) and 
that are "fixably compatible" (i.e. w/some trade-offs we could manage 
to get a language people like with no major paradoxical situation)

seems to me, therefore, that until some strawman language proposals 
come along, and Frank has agreed to start this process, there is no 
way we can discuss the specifics of layering -- we can discuss the 
philosophy of how we see languages fitting together (and your 
document was a great start on that discussion), and we can discuss 
practical implications with respect to syntax once we have some 
proposed language features to "test" them against.

Note, this still argues for exactly the sort of coevolution and 
control that Peter argues for quite eloquently in his message - but I 
simply want to start at step 2 since I feel like we've spent four 
months thinking about

At 8:00 AM -0500 3/1/02, Peter F. Patel-Schneider wrote:
>
>1/ Some of the basic notions of OWL are determined, such as the general way
>    its syntax is going to look and how it layers on top of RDF(S).

but trying to do this not anchored from language features seems odd 
to me at this point in the process - making an a priori decision 
w/respect to the layering without now taking a stab at defining what 
we'd LIKE to accomplish is just as wrong as would be taking that stab 
without having been informed by the document on layering would have 
been.

In short, I see a chicken and an egg here, and the only way I've ever 
seen chicken and egg problems cracked is to role up ones sleeves, get 
some stuff on the table from multiple perspectives, and then work to 
reconcile the views - and that's what I'm thinking we need to do 
here.  In fact, that's how the Joint Committee that created DAML+OIL 
was formed and worked, and it seems to have worked out pretty well in 
the end...


-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
AV Williams Building, Univ of Maryland		  College Park, MD 20742
http://www.cs.umd.edu/users/hendler

Received on Friday, 1 March 2002 10:22:40 UTC