[Pellet-users] Individuals don't get inferred to belong to supertypes properly

[Matthew Horridge]

Hi Tony,

I think Pellet is behaving correctly and FaCT++ might not be......

Protege 4 asks for the direct instances of a class in order to  
populate the view that you're talking about.  In the example that you  
give, the individuals aren't direct instances of AUB, which is why you  
don't see them.  I confirmed this by changing the Protege 4 code to  
show direct and indirect instances of a class and both Pellet and FaCT+ 
+ produce the same view.

So, the bug is with FaCT++ (or rather the Java bridge to FaCT++ -- the  
native code does what it's asked to do :) ).  Perhaps I should change  
Protege 4 to show indirect instances in the particular view that  
you're talking about, or at least make it optional.

If using Pellet programmatically via the OWL API then make sure you  
set the direct instances flag to false.

Cheers,

Matthew





On 29 Nov 2007, at 11:48, Tony Burdett wrote:

> I've been seeing this problem in a few guises with some class  
> inferrences I've been trying to do recently, but I've only just been  
> able to nail it down.  It seems that pellet fails to infer  
> individuals as belong to subclasses appropriately sometimes.  I've  
> knocked up a specif example to show this.
>
> I declare a Class "A" which has a single instance "a", and then a  
> Class "B" which has a single instance "b".  Then, I declare a class  
> (A U B) which is equivalent to the union of A and B.  When I  
> classify this using Pellet in protege, I would expect to see the  
> individuals a and b inferred to belong to AuB as well as A and B  
> respectively.  This isn't the case, pellet tells me there are no  
> inferred instances of  AuB.  When I classify with Fact++, however, I  
> get the answer I expect.  Presumably, this is a bug?
>
> I've attached a sample file which contains some simple union/ 
> intersection examples that should show the problem when reasoned  
> over.  This is a bit of a killer for our project, as we need to be  
> able to get individuals that belong to an anonymous "unioned"  
> classes all the time.
>
> Thanks,
>
> Tony.
>
> -- 
> Tony Burdett
> Software Developer,
> ComparaGrid.
>
> European Bioinformatics Institute
> email: tburdett at ebi.ac.uk
> tel:   01223 494624
>
> <union-inferrence- 
> test.zip>_______________________________________________
> Pellet-users mailing list
> Pellet-users at lists.owldl.com
> http://lists.owldl.com/mailman/listinfo/pellet-users
> _______________________________________________
>
> Sponsored by Clark & Parsia, LLC http://clarkparsia.com/