W3C home > Mailing lists > Public > public-owl-dev@w3.org > October to December 2011

Problematic interesectionOf construct for determining OWL Lite dialect

From: Ondrej Zamazal <ondrej.zamazal@vse.cz>
Date: Wed, 02 Nov 2011 13:11:09 +0100
Message-ID: <4EB1335D.4050406@vse.cz>
To: public-owl-dev@w3.org, protege-owl@lists.stanford.edu
Hello,
I was inspecting what constructs are admissible in OWL Lite and I have 
encountered on strange things regarding OWL species returned by OWL 
Validator [1] and by Protégé-OWL version 3.4.7.

My main concern is about "intersectionOf" construct. Looking at [2] "OWL 
Lite allows intersections of named classes and restrictions". This means 
that

C2 subClassOf C1 and (p1 some C2)

should be still valid axiom from OWL Lite perspective. However, if you 
try "determine OWL sublanguage" in Protégé then it returns OWL DL for 
this case, which is wrong. The wrong result is also returned by OWL 
Validator [2]. But if there is "equivalentTo" instead of "subClassOf" 
then OWL Validator returns "OWL Lite" correctly while Protégé still 
returns "OWL DL".

Then, I realized that if you directly decompose "intersectionOf" 
construct into an appropriate number of subclasses/equivalences than it 
is always correct in the case of OWL Validator. In the "minimalistic" 
case the following is incorrectly classified as axiom from OWL DL using 
OWL Validator and Protege:

A subclassof (B and C)

If there are two following axioms (which is equivalent representation), 
it is then correctly classified as OWL Lite by both OWL Validator and 
Protege, i.e.:

A subclassof B
A subclassof C

In the case of Protege there are some further unclear cases even if 
"intersectionOf" is decomposed e.g.

A subclassof (p1 some C2)
A subclassof C

is correctly classified as OWL Lite. However,

A equivalentTo (p1 some C2)
A equivalentTo C

is wrongly classified as OWL DL by Protege.

I am aware that now we are in the age of OWL 2 so maybe these notes are 
already obsolete, however I wanted to share with you about those unclear 
OWL dialect results. But it might also be the case that  I missed some 
point?

Thanks
Regards,
Ondrej


[1] http://www.mygrid.org.uk/OWL/Validator
[2] http://www.w3.org/TR/2004/REC-owl-features-20040210/#intersectionOf

-- 
Ondrej Svab-Zamazal, Ph.D.
University of Economics, Prague
Faculty of Informatics and Statistics
Department of Information and Knowledge Engineering
ondrej.zamazal@vse.cz http://nb.vse.cz/~svabo
Received on Thursday, 3 November 2011 17:53:07 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 3 January 2012 18:13:41 GMT