Re: Fragments discussion, continued

Hi Jim,

I sent a note [1] yesterday explaining why IBM?s position on fragment was 
close to Alan?s proposal. I will reiterate two important points justifying 
our support for a fragment that would provide scalable and complete query 
answering over large ABoxes. 

First, more enterprise data resides in traditional RDBMS than in RDF/OWL 
triple stores, and our customers rely on the plethora of critical features 
available in RDBMS (robust security, transactions, updates, scalability 
with completeness for SQL query answering, etc).  To reduce customer?s 
resistance to the adoption of OWL technology, a clear migration path based 
on standardized technologies is necessary.  From our perspective, it could 
be achieved by allowing customers to start with annotating their data with 
concepts and relationships defined in DL-Lite type of fragment without 
impact on operational performance ? i.e. performance of existing 
applications requiring quick query response time. On the other hand, since 
the data is already annotated with semantic information ?alas inexpressive 
one-, more expressive semantic views could be defined without changing the 
data. 

Second, I share Bijan?s concern expressed in [2]: the completeness of a 
fragment is a simple way to facilitate interoperability.  As I mentioned 
at this week meeting, if we decide to give up completeness, for example as 
it seems to be the case for OWL Prime, we should at least precisely define 
various levels of conformance - or incompleteness.

Finally, here are my answers to Jim?s questions with respect to the 
?RDBMS-friendly? fragment:
 1 - Supported by a group of users: enterprise customers who have large 
among of critical business data in traditional RDBMS and see (or want to 
explore )  the value of  OWL, but they want to deploy OWL technology in a 
minimally invasive way. 
 2 - Favorable property that separates them from Full and DL: scalability 
and completeness over large ABoxes
 3 - Maximal w/respect to that property: Yes, this is a nice feature of 
the language, but I don?t see it as key.

Best regards,
Achille.

[1] http://lists.w3.org/Archives/Public/public-owl-wg/2008Feb/0127.html
[2] http://lists.w3.org/Archives/Public/public-owl-wg/2008Feb/0139.html



Alan Ruttenberg <alanruttenberg@gmail.com> 
Sent by: public-owl-wg-request@w3.org
02/21/2008 10:35 AM

To
Jim Hendler <hendler@cs.rpi.edu>
cc
Boris Motik <boris.motik@comlab.ox.ac.uk>, "'Web Ontology Language ((OWL)) 
Working Group WG'" <public-owl-wg@w3.org>
Subject
Re: Fragments discussion, continued






On Feb 21, 2008, at 10:07 AM, Jim Hendler wrote:

Alan, can you explain who it is whose been asking for that fragment (as 
separate from the Prime fragment) and what the distinguishing property 
would be?

Hello Jim,

First, OWL Lite is not proposed as a Rec track fragment, so it isn't 
properly in this list. My proposal was to have that be a note, as 
an accommodation to existing users.

Regarding DL-Lite, I'm not sure I have more to add to what I've already 
said. 

3) A fragment characterized by scalability to large numbers of instances 
(not necessarily scalable tbox) , but with strong guarantees with respect 
to completeness and consistency detection. 
..
Such a fragment  fills a hole that neither of two other fragments fill, as 
It is likely that OWL Prime will not allow existentials in  a rule head 
(following pD*), and EL++ is not as scalable. 

In addition DL-Lite is implementable in relational databases with queries 
translatable to SQL. I have heard of two academic implementations ("the 
italians" &  Jeff Pan) and a commercial implementation - Clark and 
Parsia's, and the nature of the fragment is such that it would be easily 
adoptable by relational database providers. 

Finally, it is my judgement, as a user, that strong guarantees of the 
ability to detect inconsistency and give complete answers bring high value 
to science applications.

Note that these criteria have not been suggested as those that guide the 
design of OWL Prime. Should they turn out to be achievable, and there are 
implementations that show this, then I would, of course, reconsider, as I 
am more interested in the criteria than the fragment. DL-Lite is already 
sitting in this space with 3 independent implementations, so I naturally 
suggested it.

FWIW, I will be a customer for a fragment like (3) in parts of my work, 
but to be honest, I see all three as being tools that I could put to good 
use.

In my view, the argument for moving from 3 to 2 rec track fragments is not 
compelling, and will meet with resistance for  small gain. In the spirit 
of compromise, and in the interest of moving forward, I would prefer if we 
now focus on the work of answering the remaining questions we need to in 
order to sanity check and ensure ourselves that these fragments pass 
muster.  Of the 3 rec track fragments, there were significant and 
substantive questions about the details about the details OWL Prime, and 
DL-Lite. For OWL-Lite, we have some verification of the conditions that we 
set out.

Regards,
Alan