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

Re: new topic: Should Closed World Assumption(CWA) and Open World Assumption(OWA) be integrated?

From: Enrico Franconi <franconi@inf.unibz.it>
Date: Mon, 22 Aug 2011 18:08:37 +0200
Cc: Pascal Hitzler <pascal.hitzler@wright.edu>, public-owl-dev@w3.org
Message-Id: <6BADAB0C-5357-4A2D-80F4-933BE94C482B@inf.unibz.it>
To: Bijan Parsia <bparsia@cs.man.ac.uk>
I'd call LCWA the ability to specify in a standard OWA environment some predicates to behave with a CWA - for some precise notion of CWA. Roughly speaking, in my definition a predicate "is closed" if its extension is perfectly known an we can list its exact finite set of instances/tuples; i.e., it corresponds to a DB table.
By updating any of these LCWA predicates (which I call DBox predicates) we get a nonmonotonic behaviour; while if we update the ABox we remain monotonic.
Standard nonLCWA is the classical CWA, namely: whatever can not be entailed is actually true. Note that for most expressive interesting logic, this definition leads quickly to inconsistency, therefore it is not an interesting definition (see GCWA, etc, etc, from the 80'ies).
--e.

On 22 Aug 2011, at 16:54, Bijan Parsia wrote:

> On 22 Aug 2011, at 15:32, Enrico Franconi wrote:
> 
>> I guess you forgot my favourite reference on Local CWA :-)
>> 
>> InanÁ Seylan, Enrico Franconi, Jos de Bruijn: Effective Query Rewriting with Ontologies over DBoxes. IJCAI 2009: 923-925.
>> http://ijcai.org/papers09/Papers/IJCAI09-157.pdf
> 
> Ok, that causes me to reiterate my question (since I followed the Etzioni et al reference): What *isn't* "Local" CWA?
> 
> Presumably, it's not about arbitrary non-monotonic features (i.e., specially must be CWAesque reasoning).
> 
> Presumably it's stronger that merely combining OWA and CWA. (I.e., is the use of the K operator necessarily an instance of local CWA? Similarly with the use of neg?) It has to make an *assumption* right? So somehow be not specific to particular queries or axioms?
> 
> Thus, DBoxes would be exactly such, as you specify a set of predicates which are closed and thus to which CWA can apply. (I expect you could encode it in most logics with the requisite nonmonotonic features.)
> 
> Am I close? Is there a precise distinction?
> 
> Cheers,
> Bijan.
Received on Monday, 22 August 2011 16:09:15 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:59 GMT