Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-owl] How are Quantifiers / Cardinalities enforced by the Pellet Reasoner

Amrit T amrit.lists at gmail.com
Thu May 22 14:56:52 PDT 2008


Hi All,
Thanks for your reply Thomas. I can sort of see what you're saying. Here are
the things I am still not clear about.
1. I asserted in the ontology that 'A', 'B' and 'C' are disjoint classes.
Doesnt that mean that the individuals will be of a single type of those
classes? (in which case they should be distinct)
2. A's "necessary and sufficient" definition states that it should have
"some" from 'B' and 'C' classes for the property named 'property'. In the
ontology given below I have only attached it to B's instance. Do I need to
assert (due to open world assumption) that it does not have a property with
'C' explicitly to generate and inconsistency here? How do I do this at the
instance level to generate some form of inconsistency. The reason I want to
do this is because the "design" and "instance" ontologies are separate and I
need some way of knowing that the "instance" ontology's individual did not
follow its class's definition in the "design".
3. Same as above if I state that a class has minCardinality 1 for a
property, just because it has an instance which does not have the property
asserted, does not mean that it doesn't exist. In order to generate an
inconsistency here, do I need to explicitly assert this. How do I do this
this (close the system where if I state that minCardinality is 1, every
instance should have at least one of that property else there is some sort
of error flagging?)
Thanks for the help
--A.T.

On Thu, May 22, 2008 at 5:32 PM, Thomas Russ <tar at isi.edu> wrote:

>
> On May 22, 2008, at 11:47 AM, Amrit T wrote:
>
> > Hi,
> > I am trying to enforce quantifier / cardinality restriction in a
> > simple test ontology <given below>.
> > a. With the quantifiers an "existential" quantifier on the property
> > requiring someValueFrom both classes 'B' and 'C' is enforced here.
> > However in the instance I attach A's individual to only B's
> > individual. The resoner doesnt point this out as an inconsistency.
> > Why is that?
>
> Because you are not thinking of the restrictions in the proper OWL
> way.  They are not just constraints, but also INFERENCE RULES.  So,
> unless you actually have some way of proving that B's individual
> CANNOT logically be of type C, then Pellet will in fact, infer that
> the individual belongs to class C as well as B.  The information that
> you provide can be used for inference, and that is what the system does.
>
> > Also if I dont use this property to attach to any of the
> > individuals, I only get a red box but still no reasoner
> > inconsistency by doing "Check Consistency".
>
> That is because of the open world reasoning assumption.  Just because
> a value is not asserted in the ontology, it doesn't mean that the
> assertion is false.  So OWL is quite happy knowing that there is a
> filler of the property (which is an inference from the existential
> restriction), even if it doesn't happen to know which particular
> instance is that filler.  One of the expressive power benefits of OWL
> is that the reasoner can operate with such vague knowledge that there
> is a filler.  It doesn't have to be grounded in a concrete individual.
> >
> > b. Cardinality restriction of say at least one such relation are not
> > detected by the reasoner as an inconsistency if there are no such
> > relations. Again what I get is a red box surrounding the widget.
> > I am sure there is something wrong in my approach. Please advise.
>
> See above.  This is the designed behavior of OWL.  The red box is just
> a user interface aid and legacy of the Protege frames language.
> Consider it a reminder that a filler is present, but unknown.  It is
> NOT an error or inconsistency.
>
> To get an inconsistency, an OWL reasoner has to be able to prove that
> no such individual filler CAN exist.  That is much harder to do than
> just not knowing what the filler is.
>
>
> >
> > Thanks,
> > --A.T.
>
> _______________________________________________
> protege-owl mailing list
> protege-owl at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/protege-owl/attachments/20080522/2ffe7cbb/attachment.html>


More information about the protege-owl mailing list