Re: OWL DL and OWL Full

From: ewallace@cme.nist.gov
Subject: Re: OWL DL and OWL Full
Date: Mon, 29 Oct 2007 16:06:26 -0500 (EST)

> Peter F. Patel-Schneider wrote:
> 
> >Do you have anything you can point to so that the situation with respect
> >to the startups and Oracle can be analyzed?  Even the information in the
> >Oracle white paper
> >(http://www.oracle.com/technology/tech/semantic_technologies/pdf/semantic11g_dataint_twp.pdf)
> >is extremely sketchy.  About the only thing that they say is that they
> >do "efficient and scalable inferencing using major subsets of OWL".
> >
> >peter
> 
> I was on the line for Alan Wu's recent web talk on this for the Ontolog
> Forum.  On slide 7 of [1] there is a list of OWL language constructs in
> OWLPrime (the subset of OWL that 11G supports).  Alan Wu is now a member
> of this working group so I expect more details will be forth coming.
> 
> - Evan
> 
> [1] http://ontolog.cim3.net/file/work/DatabaseAndOntology/2007-10-18_AlanWu/RDBMS-RDFS-OWL-InferenceEngine--AlanWu_20071018.pdf
> 

Interesting.

With OWLPrime they support:

* rdfs:subClassOf, subPropertyOf, domain, range
* owl:TransitiveProperty, SymmetricProperty, FunctionalProperty, InverseFunctionalProperty,
* owl:inverseOf, sameAs, differentFrom
* owl:disjointWith, complementOf,
* owl:hasValue, allValuesFrom, someValuesFrom
* owl:equivalentClass, equivalentProperty

which is OWL DL minus number restrictions, disjunction, conjunction, and
instance(!).

However, repeated use of owl:hasValue puts conjunction back in.
Further, owl:complementOf can be used to turn disjunction into
conjunction.  It thus appears to me that they thus have at least the DL
SHOI.   

There is a paper on SHOI in DL'04 by Hladik and Model
(http://sunsite.informatik.rwth-aachen.de/Publications/CEUR-WS/Vol-104/22Hladik-final.pdf)
which gives a tableau system for SHOI, and shows that reasoning in it is
EXP-TIME complete.

I wonder how Oracle uses forward-chaining rules for owl:someValuesFrom.
It seems to me that this would quickly produce non-terminating
computations.

peter

Received on Tuesday, 30 October 2007 08:09:44 UTC