owl:allValuesFrom inference

[Bernhard Schandl]

Hi,

I have a question regarding OWL DL inference. Given the following axiom,

    ex:ThingMadeByMan
       a owl:Restriction ;
       owl:onProperty ex:madeBy ;
       owl:allValuesFrom ex:Man ;
    .

and the following assertions,

    ex:Bart a ex:Man .
    ex:Something ex:madeby ex:Bart .

is there a way to enfore a reasoner to infer the following?

    ex:Something a ex:ThingMadeByMan .


I was not able to produce this triple using the Pellet reasoner, and I  
would be thankful for any hints. Probably I am missing some  
constraints that I should add to the tbox?

Best,
Bernhard

[Uli Sattler]

On 25 Jun 2009, at 10:38, Bernhard Schandl wrote:

> Hi,
>

Hi,

> I have a question regarding OWL DL inference. Given the following  
> axiom,
>
>   ex:ThingMadeByMan
>      a owl:Restriction ;
>      owl:onProperty ex:madeBy ;
>      owl:allValuesFrom ex:Man ;
>   .

so, you say that, if  a ThingMadeByMan is madeBy something, then this  
something is a Man.

>
>
> and the following assertions,
>
>   ex:Bart a ex:Man .
>   ex:Something ex:madeby ex:Bart .

now you have a Man, Bart, and Something which is madeBy Bart.

>
>
> is there a way to enfore a reasoner to infer the following?
>
>   ex:Something a ex:ThingMadeByMan .
>

yes there is -- but Pellet is right in not inferring this triple from  
what you have said above because it doesn't follow from it.   What you  
want to say is that *if something is madeby a Man (and possibly by  
some other things), then this something is a ThingMadeByMan. So there  
direction of the implication needs to go the other way round an you  
need existential (someValues) restriction... in Manchester Syntax:

  Class: ThingMadeByMan
    EquivalentTo:
      madeBy some Man

This axiom together with your 2 assertions above about Bart and  
Something should then imply that Something is ThingMadeByMan

Cheers, Uli

>
> I was not able to produce this triple using the Pellet reasoner, and  
> I would be thankful for any hints. Probably I am missing some  
> constraints that I should add to the tbox?
>
> Best,
> Bernhard
>
>
>
>

[Bernhard Schandl]

Hi,

> What you want to say is that *if something is madeby a Man (and  
> possibly by some other things), then this something is a  
> ThingMadeByMan.

exactly.

> So there direction of the implication needs to go the other way  
> round an you need existential (someValues) restriction... in  
> Manchester Syntax:
>
> Class: ThingMadeByMan
>   EquivalentTo:
>     madeBy some Man

I'm not too familiar with Manchester Syntax, is this equivalent to (n3):

ex:ThingMadeByMan
    a owl:Restriction ;
    owl:onProperty ex:madeBy ;
    owl:someValuesFrom ex:Man ;
.

... because I tried this one, but stil the implication

> This axiom together with your 2 assertions above about Bart and  
> Something should then imply that Something is ThingMadeByMan

is not derived by Pellet. :-(

Also I wonder what a reasoner can actually infer from  
owl:someValuesFrom -- as far as I can tell from the spec [1] it can  
actually only be used to check the consistency of a model, but not to  
infer new facts, since the reasoner cannot decide which of the  
(possibly many) values of the property is an instance of the specified  
class. Am I missing something here?

Best, Bernhard

[Sean Bechhofer-2]

On 25 Jun 2009, at 11:12, Bernhard Schandl wrote:

> Hi,
>
>> What you want to say is that *if something is madeby a Man (and  
>> possibly by some other things), then this something is a  
>> ThingMadeByMan.
>
> exactly.
>
>> So there direction of the implication needs to go the other way  
>> round an you need existential (someValues) restriction... in  
>> Manchester Syntax:
>>
>> Class: ThingMadeByMan
>>   EquivalentTo:
>>     madeBy some Man
>
> I'm not too familiar with Manchester Syntax, is this equivalent to  
> (n3):
>
> ex:ThingMadeByMan
>    a owl:Restriction ;
>    owl:onProperty ex:madeBy ;
>    owl:someValuesFrom ex:Man ;
> .

No. The N3 above is only asserting *subclass* here. In M/cr syntax,  
it's saying:

Class: ThingMadeByMan
   SubClassOf:
     madeBy some Man

So any instance of ThingMadeByMan must be madeBy some Man, but there  
could be things which are madeBy some Man which are *not* instances  
of ThingMadeByMan. So the inference that you want to draw doesn't  
follow (and the reasoner is behaving itself :-).

        Sean

--
Sean Bechhofer
School of Computer Science
University of Manchester
sean.bechhofer@...
http://www.cs.manchester.ac.uk/people/bechhofer

[Uli Sattler]

On 25 Jun 2009, at 11:12, Bernhard Schandl wrote:

> Hi,
>
>> What you want to say is that *if something is madeby a Man (and  
>> possibly by some other things), then this something is a  
>> ThingMadeByMan.
>
> exactly.
>
>> So there direction of the implication needs to go the other way  
>> round an you need existential (someValues) restriction... in  
>> Manchester Syntax:
>>
>> Class: ThingMadeByMan
>>  EquivalentTo:
>>    madeBy some Man
>
> I'm not too familiar with Manchester Syntax, is this equivalent to  
> (n3):
>
> ex:ThingMadeByMan
>   a owl:Restriction ;
>   owl:onProperty ex:madeBy ;
>   owl:someValuesFrom ex:Man ;
> .
>
> ... because I tried this one, but stil the implication
>

hm, I'm not familiar with n3 syntax -- but i know that, in OWL, we  
have both "SubClassOf" and "EquivalentClass" statements -- and that it  
is important to distinguish the two...namely, if you say that X and Y  
are equivalent classes, then this has the same consequences as saying  
that X is a subclass of Y and Y is a subclass of X.

Now, in your example, you want to infer *from* something being made by  
men that something is a ThingMadeByMan...for which you need the  
implication 'from right to left'...or in both directions.

>> This axiom together with your 2 assertions above about Bart and  
>> Something should then imply that Something is ThingMadeByMan
>
> is not derived by Pellet. :-(
>
> Also I wonder what a reasoner can actually infer from  
> owl:someValuesFrom -- as far as I can tell from the spec [1] it can  
> actually only be used to check the consistency of a model, but not  
> to infer new facts, since the reasoner cannot decide which of the  
> (possibly many) values of the property is an instance of the  
> specified class.

I am not sure where this impression came from -- but its wrong, you  
can infer new facts: did you try your example? In your example, you  
have *stated* that Bart is a Man and that Something is madeby Bart;  
hence we (and the reasoner, too) can infer that Something is madeBy a  
Man and thus, if we also defined (!) things madeBy a Man to be    
ThingMadeByMan, then we (and the reasoner) can infer that Something is  
a ThingMadeByMan.

Cheers, Uli


> Am I missing something here?

> Best, Bernhard
>

[Irene Celino]

Some parts of this message have been removed. Learn more about Nabble's security policy.

Bernhard,
I think that what you need in N3 syntax should look like the following:

ex:ThingMadeByMan
 a owl:Restriction ;
 owl:equivalentClass [
   owl:onProperty ex:madeBy ;
   owl:allValuesFrom ex:Man ] .

Cheers,
Irene

Irene Celino

CEFRIEL - ICT Institute Politecnico di Milano

Via Fucini, 2 - 20133 Milano (Italy)

phone: [phone]

fax:   [phone]

email: Irene.Celino@...

web:   http://www.cefriel.it, http://swa.cefriel.it

 

Looking for a service? Try Service-Finder at http://demo.service-finder.eu!

 

 

2009/6/25 Uli Sattler <sattler@...>

On 25 Jun 2009, at 11:12, Bernhard Schandl wrote:

Hi,

What you want to say is that *if something is madeby a Man (and possibly by some other things), then this something is a ThingMadeByMan.

exactly.

So there direction of the implication needs to go the other way round an you need existential (someValues) restriction... in Manchester Syntax:

Class: ThingMadeByMan
 EquivalentTo:
  madeBy some Man

I'm not too familiar with Manchester Syntax, is this equivalent to (n3):

ex:ThingMadeByMan
 a owl:Restriction ;
 owl:onProperty ex:madeBy ;
 owl:someValuesFrom ex:Man ;
.

... because I tried this one, but stil the implication

hm, I'm not familiar with n3 syntax -- but i know that, in OWL, we have both "SubClassOf" and "EquivalentClass" statements -- and that it is important to distinguish the two...namely, if you say that X and Y are equivalent classes, then this has the same consequences as saying that X is a subclass of Y and Y is a subclass of X.

Now, in your example, you want to infer *from* something being made by men that something is a ThingMadeByMan...for which you need the implication 'from right to left'...or in both directions.

This axiom together with your 2 assertions above about Bart and Something should then imply that Something is ThingMadeByMan

is not derived by Pellet. :-(

Also I wonder what a reasoner can actually infer from owl:someValuesFrom -- as far as I can tell from the spec [1] it can actually only be used to check the consistency of a model, but not to infer new facts, since the reasoner cannot decide which of the (possibly many) values of the property is an instance of the specified class.

I am not sure where this impression came from -- but its wrong, you can infer new facts: did you try your example? In your example, you have *stated* that Bart is a Man and that Something is madeby Bart; hence we (and the reasoner, too) can infer that Something is madeBy a Man and thus, if we also defined (!) things madeBy a Man to be   ThingMadeByMan, then we (and the reasoner) can infer that Something is a ThingMadeByMan.

Cheers, Uli

Am I missing something here?

Best, Bernhard

[Sean Bechhofer-2]

On 25 Jun 2009, at 12:30, Sean Bechhofer wrote:

>
> On 25 Jun 2009, at 11:12, Bernhard Schandl wrote:
>
>> Hi,
>>
>>> What you want to say is that *if something is madeby a Man (and  
>>> possibly by some other things), then this something is a  
>>> ThingMadeByMan.
>>
>> exactly.
>>
>>> So there direction of the implication needs to go the other way  
>>> round an you need existential (someValues) restriction... in  
>>> Manchester Syntax:
>>>
>>> Class: ThingMadeByMan
>>>   EquivalentTo:
>>>     madeBy some Man
>>
>> I'm not too familiar with Manchester Syntax, is this equivalent to  
>> (n3):
>>
>> ex:ThingMadeByMan
>>    a owl:Restriction ;
>>    owl:onProperty ex:madeBy ;
>>    owl:someValuesFrom ex:Man ;
>> .
>
> No. The N3 above is only asserting *subclass* here.

<cough> Someone (thanks Peter :-) has pointed out to me that I'm  
wrong here. The triples above create an owl:Restriction, and then  
give it a name, which is not quite how I interpreted it. Note that  
this actually pushes you into OWL Full, as you're not allowed to name  
a restrictions in OWL DL.

> In M/cr syntax, it's saying:
>
> Class: ThingMadeByMan
>   SubClassOf:
>     madeBy some Man


The counterpart to the M/cr syntax I gave above would actually be  
something like:

ex:ThingMadeByMan
    owl:equivalentClass [
    a owl:Restriction ;
    owl:onProperty ex:madeBy ;
    owl:someValuesFrom ex:Man.].

e.g. stating that ThingMadeByMan is equivalent to the appropriate  
restriction.

As an aside, this also illustrates how easy it is (well, for me at  
least, anyway :-) to get confused by OWL represented as triples. I'd  
say it's much easier to see what's going on in an OWL (DL) model if  
it's presented using something like M/cr syntax.

Cheers,

        Sean

--
Sean Bechhofer
School of Computer Science
University of Manchester
sean.bechhofer@...
http://www.cs.manchester.ac.uk/people/bechhofer

[Pat Hayes]

On Jun 25, 2009, at 4:38 AM, Bernhard Schandl wrote:

> Hi,
>
> I have a question regarding OWL DL inference. Given the following  
> axiom,
>
>  ex:ThingMadeByMan
>     a owl:Restriction ;
>     owl:onProperty ex:madeBy ;
>     owl:allValuesFrom ex:Man ;
>  .
>
> and the following assertions,
>
>  ex:Bart a ex:Man .
>  ex:Something ex:madeby ex:Bart .
>
> is there a way to enfore a reasoner to infer the following?
>
>  ex:Something a ex:ThingMadeByMan .
>

That does not follow from what you have written. Here's a  
counterexample:

Something madeBy Bart
Something madeBy Jane

with Jane not in the class Man. This satisfies all your assumptions,  
but Something is not in the class ThingMadeByMan.

There are several ways you could extend your axioms to fix this. One  
way would be to rule out Janes by saying that madeBy was functional,  
It would then follow, in this example, that Jane sameAs Bart. Or you  
could change allValuesFrom to someValuesFrom in your definition of  
ThingMadeByMan.

Pat Hayes

>
> I was not able to produce this triple using the Pellet reasoner, and  
> I would be thankful for any hints. Probably I am missing some  
> constraints that I should add to the tbox?
>
> Best,
> Bernhard
>
>
>
>
>
>

------------------------------------------------------------
IHMC                                     [phone] or [phone]
40 South Alcaniz St.           [phone]   office
Pensacola                            [phone]   fax
FL 32502                              [phone]   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

[Rinke Hoekstra-4]

Hi,

Let me take this opportunity to point (again) to Sean's excellent set  
of OWL examples at [1].

Really helpful if you're new to/confused by OWL DL stuff.

-Rinke

[1] http://owl.man.ac.uk/2005/07/sssw/


On 25 jun 2009, at 15:56, Sean Bechhofer wrote:

>
> On 25 Jun 2009, at 12:30, Sean Bechhofer wrote:
>
>>
>> On 25 Jun 2009, at 11:12, Bernhard Schandl wrote:
>>
>>> Hi,
>>>
>>>> What you want to say is that *if something is madeby a Man (and  
>>>> possibly by some other things), then this something is a  
>>>> ThingMadeByMan.
>>>
>>> exactly.
>>>
>>>> So there direction of the implication needs to go the other way  
>>>> round an you need existential (someValues) restriction... in  
>>>> Manchester Syntax:
>>>>
>>>> Class: ThingMadeByMan
>>>>  EquivalentTo:
>>>>    madeBy some Man
>>>
>>> I'm not too familiar with Manchester Syntax, is this equivalent to  
>>> (n3):
>>>
>>> ex:ThingMadeByMan
>>>   a owl:Restriction ;
>>>   owl:onProperty ex:madeBy ;
>>>   owl:someValuesFrom ex:Man ;
>>> .
>>
>> No. The N3 above is only asserting *subclass* here.
>
> <cough> Someone (thanks Peter :-) has pointed out to me that I'm  
> wrong here. The triples above create an owl:Restriction, and then  
> give it a name, which is not quite how I interpreted it. Note that  
> this actually pushes you into OWL Full, as you're not allowed to  
> name a restrictions in OWL DL.
>
>> In M/cr syntax, it's saying:
>>
>> Class: ThingMadeByMan
>>  SubClassOf:
>>    madeBy some Man
>
>
> The counterpart to the M/cr syntax I gave above would actually be  
> something like:
>
> ex:ThingMadeByMan
>   owl:equivalentClass [
>   a owl:Restriction ;
>   owl:onProperty ex:madeBy ;
>   owl:someValuesFrom ex:Man.].
>
> e.g. stating that ThingMadeByMan is equivalent to the appropriate  
> restriction.
>
> As an aside, this also illustrates how easy it is (well, for me at  
> least, anyway :-) to get confused by OWL represented as triples. I'd  
> say it's much easier to see what's going on in an OWL (DL) model if  
> it's presented using something like M/cr syntax.
>
> Cheers,
>
> Sean
>
> --
> Sean Bechhofer
> School of Computer Science
> University of Manchester
> sean.bechhofer@...
> http://www.cs.manchester.ac.uk/people/bechhofer
>
>
>

---
Drs Rinke Hoekstra

Leibniz Center for Law      |  AI Department
Faculty of Law              |  Faculty of Sciences
Universiteit van Amsterdam  |  Vrije Universiteit
Kloveniersburgwal 48        |  De Boelelaan 1081a
1012 CX  Amsterdam          |  1081 HV Amsterdam
[phone]           |  [phone]
hoekstra@...             |  hoekstra@...

Homepage: http://www.leibnizcenter.org/users/rinke

[Sean Bechhofer-2]

On 29 Jun 2009, at 12:53, Rinke Hoekstra wrote:

> Hi,
>
> Let me take this opportunity to point (again) to Sean's excellent  
> set of OWL examples at [1].
>
> Really helpful if you're new to/confused by OWL DL stuff.
>
> -Rinke
>
> [1] http://owl.man.ac.uk/2005/07/sssw/

Thanks for the plug Rinke! There is also a slightly updated version  
[2]. Note that is all OWL 1 -- I will hopefully get round to including  
some OWL 2 content......

And not a triple in sight :-)

        Sean

[2] http://owl.cs.manchester.ac.uk/2008/07/sssw/

> On 25 jun 2009, at 15:56, Sean Bechhofer wrote:
>
>>
>> On 25 Jun 2009, at 12:30, Sean Bechhofer wrote:
>>
>>>
>>> On 25 Jun 2009, at 11:12, Bernhard Schandl wrote:
>>>
>>>> Hi,
>>>>
>>>>> What you want to say is that *if something is madeby a Man (and  
>>>>> possibly by some other things), then this something is a  
>>>>> ThingMadeByMan.
>>>>
>>>> exactly.
>>>>
>>>>> So there direction of the implication needs to go the other way  
>>>>> round an you need existential (someValues) restriction... in  
>>>>> Manchester Syntax:
>>>>>
>>>>> Class: ThingMadeByMan
>>>>> EquivalentTo:
>>>>>   madeBy some Man
>>>>
>>>> I'm not too familiar with Manchester Syntax, is this equivalent  
>>>> to (n3):
>>>>
>>>> ex:ThingMadeByMan
>>>>  a owl:Restriction ;
>>>>  owl:onProperty ex:madeBy ;
>>>>  owl:someValuesFrom ex:Man ;
>>>> .
>>>
>>> No. The N3 above is only asserting *subclass* here.
>>
>> <cough> Someone (thanks Peter :-) has pointed out to me that I'm  
>> wrong here. The triples above create an owl:Restriction, and then  
>> give it a name, which is not quite how I interpreted it. Note that  
>> this actually pushes you into OWL Full, as you're not allowed to  
>> name a restrictions in OWL DL.
>>
>>> In M/cr syntax, it's saying:
>>>
>>> Class: ThingMadeByMan
>>> SubClassOf:
>>>   madeBy some Man
>>
>>
>> The counterpart to the M/cr syntax I gave above would actually be  
>> something like:
>>
>> ex:ThingMadeByMan
>>  owl:equivalentClass [
>>  a owl:Restriction ;
>>  owl:onProperty ex:madeBy ;
>>  owl:someValuesFrom ex:Man.].
>>
>> e.g. stating that ThingMadeByMan is equivalent to the appropriate  
>> restriction.
>>
>> As an aside, this also illustrates how easy it is (well, for me at  
>> least, anyway :-) to get confused by OWL represented as triples.  
>> I'd say it's much easier to see what's going on in an OWL (DL)  
>> model if it's presented using something like M/cr syntax.
>>
>> Cheers,
>>
>> Sean
>>
>> --
>> Sean Bechhofer
>> School of Computer Science
>> University of Manchester
>> sean.bechhofer@...
>> http://www.cs.manchester.ac.uk/people/bechhofer
>>
>>
>>
>
>
>
> ---
> Drs Rinke Hoekstra
>
> Leibniz Center for Law      |  AI Department
> Faculty of Law              |  Faculty of Sciences
> Universiteit van Amsterdam  |  Vrije Universiteit
> Kloveniersburgwal 48        |  De Boelelaan 1081a
> 1012 CX  Amsterdam          |  1081 HV Amsterdam
> [phone]           |  [phone]
> hoekstra@...             |  hoekstra@...
>
> Homepage: http://www.leibnizcenter.org/users/rinke
>
>
>
>
>
>

--
Sean Bechhofer
School of Computer Science
University of Manchester
sean.bechhofer@...
http://www.cs.manchester.ac.uk/people/bechhofer

[Sean Bechhofer-2]

On 29 Jun 2009, at 12:53, Rinke Hoekstra wrote:

> Hi,
>
> Let me take this opportunity to point (again) to Sean's excellent  
> set of OWL examples at [1].
>
> Really helpful if you're new to/confused by OWL DL stuff.
>
> -Rinke
>
> [1] http://owl.man.ac.uk/2005/07/sssw/

This year's new, fresh, exciting examples now available!

http://owl.cs.manchester.ac.uk/2009/07/sssw

        Sean

--
Sean Bechhofer
School of Computer Science
University of Manchester
sean.bechhofer@...
http://www.cs.manchester.ac.uk/people/bechhofer