Problem on reasoning based on cardinality with simulating closed world assumption

["Kang, Sungku" <sk...@illinois.edu>]

Hello,

I'm working on the project which requires that the number of object related to a subject (cardinality) to classify instances.

So, I'm testing the reasoning based on cardinality with simulating closed world assumption.

I introduced several constraint to simulate closed world assumption.

I attached a small test owl file which satisfies the constraint.

The description of the test owl file is as below.

------------------------------------------------------

-There are two top-level classes, "Polygon" and "Side".
(To simulate closed world assumption, they are enumerated classes which don't allow possible additional instances)

-There is an object property "hasSide" relates "Polygon" with "Side".

-There are subclasses of "Polygon", "Square" and "Triangle".
 The definition of "Triangle" is "Polygon and (hasSide exactly 3 Thing)"
 The definition of "Square" is "Polygon and (hasSide exactly 4 Thing)"

-There are two instances belong to "Polygon", which are "Triangle1" and "Square1".
 There are four instances belong to "Side", from "Side1" to "Side4".
(To simulate closed world assumption, all instances are declared not to be the same.)

-"Triangle1" is related with three of the "Side"
  It should not be classified as a "Triangle" since Negative Object Property Assertion(NPA) is needed
  to explicitly declare the one "Side" which is not declared to be related with "Triangle1" is not related

-"Square1" is related with four of the "Side" (i.e., all of the "Side")
  It should be classified as a "Square".

------------------------------------------------------

However, Jena reasoner could not classify "Square1" as expected even though Protege could.

Can someone check the test owl file and let me know what the issue is?

I don't think the ontology spec is the reason, but may be I need to set up something for Jena reasoner.

Thanks,

SungKu

Re: Problem on reasoning based on cardinality with simulating closed world assumption

[Dave Reynolds <da...@gmail.com>]

This list doesn't support attachments so we can't see your ontology.

However, if your ontology works with Protege then it is probably
correct. The jena rule reasoners don't do anything useful with
cardinalities above 1. As I mentioned before you need to look at Pellet
or similar to get complete DL reasoning.

Dave

On 02/08/13 16:51, Kang, Sungku wrote:
> Hello,
> 
> I'm working on the project which requires that the number of object 
> related to a subject (cardinality) to classify instances.
> 
> So, I'm testing the reasoning based on cardinality with simulating 
> closed world assumption.
> 
> I introduced several constraint to simulate closed world assumption.
> 
> I attached a small test owl file which satisfies the constraint.
> 
> The description of the test owl file is as below.
> 
> ------------------------------------------------------
> 
> -There are two top-level classes, "Polygon" and "Side".
> (To simulate closed world assumption, they are enumerated classes which 
> don't allow possible additional instances)
> 
> -There is an object property "hasSide" relates "Polygon" with "Side".
> 
> -There are subclasses of "Polygon", "Square" and "Triangle".
>   The definition of "Triangle" is "Polygon and (hasSide exactly 3 Thing)"
>   The definition of "Square" is "Polygon and (hasSide exactly 4 Thing)"
> 
> -There are two instances belong to "Polygon", which are "Triangle1" and 
> "Square1".
>   There are four instances belong to "Side", from "Side1" to "Side4".
> (To simulate closed world assumption, all instances are declared not to 
> be the same.)
> 
> -"Triangle1" is related with three of the "Side"
>    It should not be classified as a "Triangle" since Negative Object 
> Property Assertion(NPA) is needed
>    to explicitly declare the one "Side" which is not declared to be 
> related with "Triangle1" is not related
> 
> -"Square1" is related with four of the "Side" (i.e., all of the "Side")
>    It should be classified as a "Square".
> 
> ------------------------------------------------------
> 
> However, Jena reasoner could not classify "Square1" as expected even 
> though Protege could.
> 
> Can someone check the test owl file and let me know what the issue is?
> 
> I don't think the ontology spec is the reason, but may be I need to set 
> up something for Jena reasoner.
> 
> Thanks,
> 
> SungKu