Re: Again on All humans love (all) cats

Marco

The wording of OWL 2 DL global restrictions on axioms [1], for my
understanding, does not require the "top property" to be the top in
the partial order of object properties. If we add  ObjectProperty:
pHuman  SubPropertyOf: owl:topObjectProperty, which itself is an
tautology (or as you do by defining the property "identity"), would
lead to an OWL 2 Full ontology as it violates the restriction on the
property hierarchy.

Whether this is a bug in OWL 2 DL is debatable. I found that, in the
SROIQ paper [2], the regularity requirement is also defined on axioms
(thus, syntax specifications), not on inferred role inclusions.

[1] http://www.w3.org/TR/owl2-syntax/#Global_Restrictions_on_Axioms_in_OWL_2_DL
[2] http://www.cs.man.ac.uk/~sattler/publications/sroiq-TR.pdf

Cheers
Jie

On Wed, Oct 13, 2010 at 04:06, Marco Colombetti <colombet@elet.polimi.it> wrote:
> I may have missed some message in this thread, but I still do not understand
> whether
>
>  pHuman o topObjectProperty o pCat SubPropertyOf loves
>
> is acceptable in OWL 2 DL, or it isn't because it brings in a circular
> dependency between loves and topObjectProperty.
>
> I made a few experiments with existing reasoners: the chain subproperty
> axiom works fine with HermiT (I mean it infers all assertions of form
> loves(human,cat)), does not work with Pellet (no assertion of form
> loves(human,cat) is inferred), and it makes my Fact++ crash.
>
> Best,
>
> Marco
>
> PS. Here is my ontology:
>
> Class: Human
>    EquivalentTo: pHuman some owl:Thing
> Class: Cat
>    EquivalentTo: pCat some owl:Thing
> ObjectProperty: identity
>    SubPropertyOf: owl:topObjectProperty
>    Characteristics: Functional, Reflexive
> ObjectProperty: pHuman
>    SubPropertyOf: identity
> ObjectProperty: pCat
>    SubPropertyOf: identity
> ObjectProperty: loves
>    SubPropertyOf: owl:topObjectProperty
>    SubPropertyChain: pHuman o owl:topObjectProperty o pCat
>    Domain: Human
>    Range: Cat
>
>
>

Received on Wednesday, 13 October 2010 15:35:35 UTC