Property union

classic Classic list List threaded Threaded
26 messages Options
12
Reply | Threaded
Open this post in threaded view
|

Property union

Aqualung
Here's an OWL 2 question: can I say that a property is the union of two (or
more) properties ? (I am, in effect, not sure if SROIQ can accommodate this,
can it?)

C

Cristian Cocos
Post Doctoral Fellow
Centre for Logic and Information, StFX University 54B St. Mary's Street,
Antigonish NS, Canada B2G 2W5
Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397

Current research: "Building Decision-Support Through Dynamic Workflow
Systems for Health Care"



_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

Timothy Redmond

I believe that the answer is that you can't do this.  While there are
several experts here who might be  able to either correct me or give a
stronger answer, I suspect that a better list for this type of question
is the owl-dev mailing list [1].

-Timothy

[1] http://lists.w3.org/Archives/Public/public-owl-dev/


On 09/20/2010 09:38 AM, Cristian Cocos wrote:

> Here's an OWL 2 question: can I say that a property is the union of two (or
> more) properties ? (I am, in effect, not sure if SROIQ can accommodate this,
> can it?)
>
> C
>
> Cristian Cocos
> Post Doctoral Fellow
> Centre for Logic and Information, StFX University 54B St. Mary's Street,
> Antigonish NS, Canada B2G 2W5
> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>
> Current research: "Building Decision-Support Through Dynamic Workflow
> Systems for Health Care"
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>    

_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

Alan Ruttenberg-2
If :a and :b are the properties that you want to union, then an  
approximation is to make the union :u the superproperty of both. This  
will break down if you either make property assertions using :u  
directly or have axioms that might entail such assertions.

-Alan

On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>  
wrote:

>
> I believe that the answer is that you can't do this.  While there  
> are several experts here who might be  able to either correct me or  
> give a stronger answer, I suspect that a better list for this type  
> of question is the owl-dev mailing list [1].
>
> -Timothy
>
> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>
>
> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>> Here's an OWL 2 question: can I say that a property is the union of  
>> two (or
>> more) properties ? (I am, in effect, not sure if SROIQ can  
>> accommodate this,
>> can it?)
>>
>> C
>>
>> Cristian Cocos
>> Post Doctoral Fellow
>> Centre for Logic and Information, StFX University 54B St. Mary's  
>> Street,
>> Antigonish NS, Canada B2G 2W5
>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>
>> Current research: "Building Decision-Support Through Dynamic Workflow
>> Systems for Health Care"
>>
>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

regmito
For exact definition, beside the subproperty axioms, you also need the following axiom:

(a some owl:Thing) or (b some owl:Thing) EquivalentTo u some owl:Thing.

And if you want inference on individual values, you also need a couple of SWRL rules:

a(?x, ?y) -> u(?x, ?y);
b(?x, ?y) -> u(?x, ?y).

For more systematic approach, property union and intersection should be allowed as property expressions that are similar to class expressions: e.g. (a or b); (a and b). However the OWL 2 specification misses this.

Igor


Dr Igor Toujilov
Senior Research Fellow
Centre for Health Informatics and Multiprofessional Education
University College London
Highgate Hill
London, UK
N19 5LW

Tel: +44 (0)20 7288 3960


________________________________________
From: [hidden email] [[hidden email]] on behalf of Alan Ruttenberg [[hidden email]]
Sent: 20 September 2010 20:11
To: Submit feedback for Protege 4.x
Cc: [hidden email]
Subject: Re: [p4-feedback] Property union

If :a and :b are the properties that you want to union, then an
approximation is to make the union :u the superproperty of both. This
will break down if you either make property assertions using :u
directly or have axioms that might entail such assertions.

-Alan

On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>
wrote:

>
> I believe that the answer is that you can't do this.  While there
> are several experts here who might be  able to either correct me or
> give a stronger answer, I suspect that a better list for this type
> of question is the owl-dev mailing list [1].
>
> -Timothy
>
> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>
>
> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>> Here's an OWL 2 question: can I say that a property is the union of
>> two (or
>> more) properties ? (I am, in effect, not sure if SROIQ can
>> accommodate this,
>> can it?)
>>
>> C
>>
>> Cristian Cocos
>> Post Doctoral Fellow
>> Centre for Logic and Information, StFX University 54B St. Mary's
>> Street,
>> Antigonish NS, Canada B2G 2W5
>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>
>> Current research: "Building Decision-Support Through Dynamic Workflow
>> Systems for Health Care"
>>
>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

Alan Ruttenberg-2
On Mon, Sep 20, 2010 at 9:57 PM, regmito <[hidden email]> wrote:
> For exact definition, beside the subproperty axioms, you also need the following axiom:
>
> (a some owl:Thing) or (b some owl:Thing) EquivalentTo u some owl:Thing.

Doesn't help. You can still have a model where one value for a and a
different and additional value for u is asserted.

topObjectProperty
 u
  a
  b

instances:
i1
i2
i3
i4

propertyassertions:

i1 a i2
i1 b i3
i1 u i4

inferred: i1 u i2,i3,i4

i1 a some thing? yes
so lhs is satisfied.
i1 u some thing? yes
so rhs satisfied

u > union a b

>
> And if you want inference on individual values, you also need a couple of SWRL rules:
>
> a(?x, ?y) -> u(?x, ?y);
> b(?x, ?y) -> u(?x, ?y).

Also doesn't help. This is exactly what the subproperty axioms do.

> For more systematic approach, property union and intersection should be allowed as property expressions that are similar to class expressions: e.g. (a or b); (a and b). However the OWL 2 specification misses this.
>
> Igor
>
>
> Dr Igor Toujilov
> Senior Research Fellow
> Centre for Health Informatics and Multiprofessional Education
> University College London
> Highgate Hill
> London, UK
> N19 5LW
>
> Tel: +44 (0)20 7288 3960
>
>
> ________________________________________
> From: [hidden email] [[hidden email]] on behalf of Alan Ruttenberg [[hidden email]]
> Sent: 20 September 2010 20:11
> To: Submit feedback for Protege 4.x
> Cc: [hidden email]
> Subject: Re: [p4-feedback] Property union
>
> If :a and :b are the properties that you want to union, then an
> approximation is to make the union :u the superproperty of both. This
> will break down if you either make property assertions using :u
> directly or have axioms that might entail such assertions.
>
> -Alan
>
> On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>
> wrote:
>
>>
>> I believe that the answer is that you can't do this.  While there
>> are several experts here who might be  able to either correct me or
>> give a stronger answer, I suspect that a better list for this type
>> of question is the owl-dev mailing list [1].
>>
>> -Timothy
>>
>> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>>
>>
>> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>>> Here's an OWL 2 question: can I say that a property is the union of
>>> two (or
>>> more) properties ? (I am, in effect, not sure if SROIQ can
>>> accommodate this,
>>> can it?)
>>>
>>> C
>>>
>>> Cristian Cocos
>>> Post Doctoral Fellow
>>> Centre for Logic and Information, StFX University 54B St. Mary's
>>> Street,
>>> Antigonish NS, Canada B2G 2W5
>>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>>
>>> Current research: "Building Decision-Support Through Dynamic Workflow
>>> Systems for Health Care"
>>>
>>>
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

regmito
Thanks, I see what you mean. But how about restricting the domain of u:

ObjectProperty: u
    Domain: D

u
  a
  b

Class: D
     SubClassOf:
        u only
            ((inverse(a) some owl:Thing)
             or (inverse(b) some owl:Thing))

-- Igor
________________________________________
From: [hidden email] [[hidden email]] on behalf of Alan Ruttenberg [[hidden email]]
Sent: 21 September 2010 06:28
To: Submit feedback for Protege 4.x
Subject: Re: [p4-feedback] Property union

On Mon, Sep 20, 2010 at 9:57 PM, regmito <[hidden email]> wrote:
> For exact definition, beside the subproperty axioms, you also need the following axiom:
>
> (a some owl:Thing) or (b some owl:Thing) EquivalentTo u some owl:Thing.

Doesn't help. You can still have a model where one value for a and a
different and additional value for u is asserted.

topObjectProperty
 u
  a
  b

instances:
i1
i2
i3
i4

propertyassertions:

i1 a i2
i1 b i3
i1 u i4

inferred: i1 u i2,i3,i4

i1 a some thing? yes
so lhs is satisfied.
i1 u some thing? yes
so rhs satisfied

u > union a b

>
> And if you want inference on individual values, you also need a couple of SWRL rules:
>
> a(?x, ?y) -> u(?x, ?y);
> b(?x, ?y) -> u(?x, ?y).

Also doesn't help. This is exactly what the subproperty axioms do.

> For more systematic approach, property union and intersection should be allowed as property expressions that are similar to class expressions: e.g. (a or b); (a and b). However the OWL 2 specification misses this.
>
> Igor
>
>
> Dr Igor Toujilov
> Senior Research Fellow
> Centre for Health Informatics and Multiprofessional Education
> University College London
> Highgate Hill
> London, UK
> N19 5LW
>
> Tel: +44 (0)20 7288 3960
>
>
> ________________________________________
> From: [hidden email] [[hidden email]] on behalf of Alan Ruttenberg [[hidden email]]
> Sent: 20 September 2010 20:11
> To: Submit feedback for Protege 4.x
> Cc: [hidden email]
> Subject: Re: [p4-feedback] Property union
>
> If :a and :b are the properties that you want to union, then an
> approximation is to make the union :u the superproperty of both. This
> will break down if you either make property assertions using :u
> directly or have axioms that might entail such assertions.
>
> -Alan
>
> On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>
> wrote:
>
>>
>> I believe that the answer is that you can't do this.  While there
>> are several experts here who might be  able to either correct me or
>> give a stronger answer, I suspect that a better list for this type
>> of question is the owl-dev mailing list [1].
>>
>> -Timothy
>>
>> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>>
>>
>> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>>> Here's an OWL 2 question: can I say that a property is the union of
>>> two (or
>>> more) properties ? (I am, in effect, not sure if SROIQ can
>>> accommodate this,
>>> can it?)
>>>
>>> C
>>>
>>> Cristian Cocos
>>> Post Doctoral Fellow
>>> Centre for Logic and Information, StFX University 54B St. Mary's
>>> Street,
>>> Antigonish NS, Canada B2G 2W5
>>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>>
>>> Current research: "Building Decision-Support Through Dynamic Workflow
>>> Systems for Health Care"
>>>
>>>
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Use of inverseFunctional for data properties

Evain, Jean-Pierre
In reply to this post by Alan Ruttenberg-2
Good morning.

While looking at the FOAF ontology on Protégé 4.0, different properties, although declared as data properties, would be displayed as data and object properties.

Would you confirm that this results from the fact that these properties are also declared as inverseFunctional, which would seem to make them look like object properties to Protégé?

It also seems, still on Protégé 4.0 that it is not proposed to declare data properties as being inverseFunctional. Any reason?

Interesting to note that reasoners accept all this.

What do you think?

Regards,

Jean-pierre
-----------------------------------------
**************************************************
This email and any files transmitted with it
are confidential and intended solely for the
use of the individual or entity to whom they
are addressed.
If you have received this email in error,
please notify the system manager.
This footnote also confirms that this email
message has been swept by the mailgateway
**************************************************

_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Thomas Schneider-5

On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:

> Good morning.
>
> While looking at the FOAF ontology on Protégé 4.0, different  
> properties, although declared as data properties, would be displayed  
> as data and object properties.
>
> Would you confirm that this results from the fact that these  
> properties are also declared as inverseFunctional, which would seem  
> to make them look like object properties to Protégé?

I don't know the FOAF ontology, but I suspect that the described  
behaviour is due to punning, which is allowed in OWL 2: two entities  
of different type (e.g. an object property and a data property) can  
have the same name. It is possible that, if the parser reads an  
InverseFunctional statement, that the property is punned and treated  
as an object property for this particular statement and as a data  
property otherwise.

However, I can't say for sure that this is the actual reason.

> It also seems, still on Protégé 4.0 that it is not proposed to  
> declare data properties as being inverseFunctional. Any reason?

Neither in 4.0 nor in 4.1 can you declare data properties as inverse  
functional. The reason is that the OWL specification doesn't allow  
this. If it were allowed, reasoning would become an undecidable problem.

> Interesting to note that reasoners accept all this.

I'm not sure what exactly you mean. If you invoked the reasoner within  
Protégé, it worked on the representation of the ontology created by  
Protégé (or the underlying OWL API). Since the properties in question  
had been punned during parsing as described above, the reasoner got  
the InverseFunctional statements for object properties. However, if  
you invoked a reasoner directly, e.g., Pellet on the command line,  
then they might simply have used punning too.

Cheers

Thomas

>
> What do you think?
>
> Regards,
>
> Jean-pierre
> -----------------------------------------
> **************************************************
> This email and any files transmitted with it
> are confidential and intended solely for the
> use of the individual or entity to whom they
> are addressed.
> If you have received this email in error,
> please notify the system manager.
> This footnote also confirms that this email
> message has been swept by the mailgateway
> **************************************************
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

PGP.sig (210 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Evain, Jean-Pierre
Thanks for the quick reply.

Got the last point on reasoning after punning.

Interesting to note that OWL (I guess inc. FULL) doesn't allow the use of inverseFunctional for data properties. I believe you ;-) but I need to check this myself to have the right arguments when explaining this. Does the spec say this specifically?

Punning in OWL-2 is interesting. We have the case (not in FOAF) where we have to cover two choices e.g. for classification scheme between an object property linking to a SKOS concept OR to a literal or string. Any link where I can find more information on this?

Apologies for asking more...

Thanks in advance

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf Of Thomas Schneider
Sent: mardi, 21. septembre 2010 10:04
To: Submit feedback for Protege 4.x
Subject: Re: [p4-feedback] Use of inverseFunctional for data properties


On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:

> Good morning.
>
> While looking at the FOAF ontology on Protégé 4.0, different
> properties, although declared as data properties, would be displayed
> as data and object properties.
>
> Would you confirm that this results from the fact that these
> properties are also declared as inverseFunctional, which would seem to
> make them look like object properties to Protégé?

I don't know the FOAF ontology, but I suspect that the described behaviour is due to punning, which is allowed in OWL 2: two entities of different type (e.g. an object property and a data property) can have the same name. It is possible that, if the parser reads an InverseFunctional statement, that the property is punned and treated as an object property for this particular statement and as a data property otherwise.

However, I can't say for sure that this is the actual reason.

> It also seems, still on Protégé 4.0 that it is not proposed to declare
> data properties as being inverseFunctional. Any reason?

Neither in 4.0 nor in 4.1 can you declare data properties as inverse functional. The reason is that the OWL specification doesn't allow this. If it were allowed, reasoning would become an undecidable problem.

> Interesting to note that reasoners accept all this.

I'm not sure what exactly you mean. If you invoked the reasoner within Protégé, it worked on the representation of the ontology created by Protégé (or the underlying OWL API). Since the properties in question had been punned during parsing as described above, the reasoner got the InverseFunctional statements for object properties. However, if you invoked a reasoner directly, e.g., Pellet on the command line, then they might simply have used punning too.

Cheers

Thomas

>
> What do you think?
>
> Regards,
>
> Jean-pierre
> -----------------------------------------
> **************************************************
> This email and any files transmitted with it are confidential and
> intended solely for the use of the individual or entity to whom they
> are addressed.
> If you have received this email in error, please notify the system
> manager.
> This footnote also confirms that this email message has been swept by
> the mailgateway
> **************************************************
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback

+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff



_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

Thomas Schneider-5
In reply to this post by regmito

On 21 Sep 2010, at 07:44, regmito wrote:

> Thanks, I see what you mean. But how about restricting the domain of  
> u:
>
> ObjectProperty: u
>    Domain: D
>
> u
>  a
>  b
>
> Class: D
>     SubClassOf:
>        u only
>            ((inverse(a) some owl:Thing)
>             or (inverse(b) some owl:Thing))
>
> -- Igor
Computer says no ... ;)

This still allows models like this:

* Individuals i,j,k,l
* a -> {(i,j), (j,i)}
* b -> {(k,l), (l,k)}
* u -> {(i,j), (j,i), (k,l), (l,k), (i,k), (k,i), (j,l), (l,j)}
* D -> {i,j,k,l}

This model satisfies the axioms

a subPropertyOf u
b subPropertyOf u
(a some owl:Thing) or (b some Thing) EquivalentTo u some Thing.
u Domain: D
D SubClassOf u only ((inverse(a) some Thing) or (inverse(b) some Thing))

I suspect that the symmetry of a,b,u in this counterexample is not  
crucial.

Cheers

Thomas

> ________________________________________
> From: [hidden email] [[hidden email]
> ] on behalf of Alan Ruttenberg [[hidden email]]
> Sent: 21 September 2010 06:28
> To: Submit feedback for Protege 4.x
> Subject: Re: [p4-feedback] Property union
>
> On Mon, Sep 20, 2010 at 9:57 PM, regmito <[hidden email]> wrote:
>> For exact definition, beside the subproperty axioms, you also need  
>> the following axiom:
>>
>> (a some owl:Thing) or (b some owl:Thing) EquivalentTo u some  
>> owl:Thing.
>
> Doesn't help. You can still have a model where one value for a and a
> different and additional value for u is asserted.
>
> topObjectProperty
> u
>  a
>  b
>
> instances:
> i1
> i2
> i3
> i4
>
> propertyassertions:
>
> i1 a i2
> i1 b i3
> i1 u i4
>
> inferred: i1 u i2,i3,i4
>
> i1 a some thing? yes
> so lhs is satisfied.
> i1 u some thing? yes
> so rhs satisfied
>
> u > union a b
>
>>
>> And if you want inference on individual values, you also need a  
>> couple of SWRL rules:
>>
>> a(?x, ?y) -> u(?x, ?y);
>> b(?x, ?y) -> u(?x, ?y).
>
> Also doesn't help. This is exactly what the subproperty axioms do.
>
>> For more systematic approach, property union and intersection  
>> should be allowed as property expressions that are similar to class  
>> expressions: e.g. (a or b); (a and b). However the OWL 2  
>> specification misses this.
>>
>> Igor
>>
>>
>> Dr Igor Toujilov
>> Senior Research Fellow
>> Centre for Health Informatics and Multiprofessional Education
>> University College London
>> Highgate Hill
>> London, UK
>> N19 5LW
>>
>> Tel: +44 (0)20 7288 3960
>>
>>
>> ________________________________________
>> From: [hidden email] [[hidden email]
>> ] on behalf of Alan Ruttenberg [[hidden email]]
>> Sent: 20 September 2010 20:11
>> To: Submit feedback for Protege 4.x
>> Cc: [hidden email]
>> Subject: Re: [p4-feedback] Property union
>>
>> If :a and :b are the properties that you want to union, then an
>> approximation is to make the union :u the superproperty of both. This
>> will break down if you either make property assertions using :u
>> directly or have axioms that might entail such assertions.
>>
>> -Alan
>>
>> On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>
>> wrote:
>>
>>>
>>> I believe that the answer is that you can't do this.  While there
>>> are several experts here who might be  able to either correct me or
>>> give a stronger answer, I suspect that a better list for this type
>>> of question is the owl-dev mailing list [1].
>>>
>>> -Timothy
>>>
>>> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>>>
>>>
>>> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>>>> Here's an OWL 2 question: can I say that a property is the union of
>>>> two (or
>>>> more) properties ? (I am, in effect, not sure if SROIQ can
>>>> accommodate this,
>>>> can it?)
>>>>
>>>> C
>>>>
>>>> Cristian Cocos
>>>> Post Doctoral Fellow
>>>> Centre for Logic and Information, StFX University 54B St. Mary's
>>>> Street,
>>>> Antigonish NS, Canada B2G 2W5
>>>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>>>
>>>> Current research: "Building Decision-Support Through Dynamic  
>>>> Workflow
>>>> Systems for Health Care"
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> p4-feedback mailing list
>>>> [hidden email]
>>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>>>
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

PGP.sig (210 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Thomas Schneider-5
In reply to this post by Evain, Jean-Pierre
Hi Jean-Pierre,

On 21 Sep 2010, at 09:20, Evain, Jean-Pierre wrote:

> Interesting to note that OWL (I guess inc. FULL) doesn't allow the  
> use of inverseFunctional for data properties. I believe you ;-) but  
> I need to check this myself to have the right arguments when  
> explaining this. Does the spec say this specifically?

I haven't found a note on this in the structural specification  
document [1]. However, my claim was based on the results in the  
technical paper [2], where the expressions allowed in "unary key  
boxes" correspond exactly to inverse-functionality constraints on  
datatype properties. Such unrestricted key boxes make reasoning  
undecidable already if added to the basic description logic ALC, which  
is a fragment of OWL. However, the same paper points out restrictions  
to key statements under which reasoning becomes decidable, but if I'm  
not mistaken, there is no complete coverage of inverse-functional DPs  
in current reasoners.

> Punning in OWL-2 is interesting. We have the case (not in FOAF)  
> where we have to cover two choices e.g. for classification scheme  
> between an object property linking to a SKOS concept OR to a literal  
> or string. Any link where I can find more information on this?

You could start here [3,4] and follow the references.

[1] http://www.w3.org/TR/2009/CR-owl2-syntax-20090611/

[2] C. Lutz, C. Areces, I. Horrocks, and U. Sattler. Keys, Nominals,  
and Concrete Domains. Journal of Artificial Intelligence Research,  
Vol. 23, pages 667-726, 2005. PDF: http://www.cs.man.ac.uk/~horrocks/Publications/download/2004/LAHS04a.pdf

[3] http://www.w3.org/TR/owl2-new-features/#F12:_Punning
[4] http://www.w3.org/2007/OWL/wiki/Punning

Cheers

Thomas

> Apologies for asking more...
>
> Thanks in advance
>
> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]
> ] On Behalf Of Thomas Schneider
> Sent: mardi, 21. septembre 2010 10:04
> To: Submit feedback for Protege 4.x
> Subject: Re: [p4-feedback] Use of inverseFunctional for data  
> properties
>
>
> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>
>> Good morning.
>>
>> While looking at the FOAF ontology on Protégé 4.0, different
>> properties, although declared as data properties, would be displayed
>> as data and object properties.
>>
>> Would you confirm that this results from the fact that these
>> properties are also declared as inverseFunctional, which would seem  
>> to
>> make them look like object properties to Protégé?
>
> I don't know the FOAF ontology, but I suspect that the described  
> behaviour is due to punning, which is allowed in OWL 2: two entities  
> of different type (e.g. an object property and a data property) can  
> have the same name. It is possible that, if the parser reads an  
> InverseFunctional statement, that the property is punned and treated  
> as an object property for this particular statement and as a data  
> property otherwise.
>
> However, I can't say for sure that this is the actual reason.
>
>> It also seems, still on Protégé 4.0 that it is not proposed to  
>> declare
>> data properties as being inverseFunctional. Any reason?
>
> Neither in 4.0 nor in 4.1 can you declare data properties as inverse  
> functional. The reason is that the OWL specification doesn't allow  
> this. If it were allowed, reasoning would become an undecidable  
> problem.
>
>> Interesting to note that reasoners accept all this.
>
> I'm not sure what exactly you mean. If you invoked the reasoner  
> within Protégé, it worked on the representation of the ontology  
> created by Protégé (or the underlying OWL API). Since the properties  
> in question had been punned during parsing as described above, the  
> reasoner got the InverseFunctional statements for object properties.  
> However, if you invoked a reasoner directly, e.g., Pellet on the  
> command line, then they might simply have used punning too.
>
> Cheers
>
> Thomas
>
>>
>> What do you think?
>>
>> Regards,
>>
>> Jean-pierre
>> -----------------------------------------
>> **************************************************
>> This email and any files transmitted with it are confidential and
>> intended solely for the use of the individual or entity to whom they
>> are addressed.
>> If you have received this email in error, please notify the system
>> manager.
>> This footnote also confirms that this email message has been swept by
>> the mailgateway
>> **************************************************
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
> +
> ----------------------------------------------------------------------+
> |  Dr Thomas Schneider                    schneider (at)  
> cs.man.ac.uk  |
> |  School of Computer Science       http://www.cs.man.ac.uk/ 
> ~schneidt  |
> |  Kilburn Building, Room 2.114                 phone +44 161  
> 2756136  |
> |  University of  
> Manchester                                            |
> |  Oxford Road                                             _///
> _       |
> |  Manchester M13 9PL                                      
> (o~o)       |
> +-----------------------------------------------------oOOO--(_)--
> OOOo--+
>
> Fremantle (vb.)
>   To steal things not worth the bother of stealing. One steals cars,
>   money and silver. Book matches, airline eyepatches and individual
>   pots of Trust House Forte apricot jam are merely fremantled.
>
>                   Douglas Adams, John Lloyd: The Deeper Meaning of  
> Liff
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

PGP.sig (210 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Evain, Jean-Pierre
Appreciate it, thanks, JP

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf Of Thomas Schneider
Sent: mardi, 21. septembre 2010 10:47
To: Submit feedback for Protege 4.x
Subject: Re: [p4-feedback] Use of inverseFunctional for data properties

Hi Jean-Pierre,

On 21 Sep 2010, at 09:20, Evain, Jean-Pierre wrote:

> Interesting to note that OWL (I guess inc. FULL) doesn't allow the use
> of inverseFunctional for data properties. I believe you ;-) but I need
> to check this myself to have the right arguments when explaining this.
> Does the spec say this specifically?

I haven't found a note on this in the structural specification document [1]. However, my claim was based on the results in the technical paper [2], where the expressions allowed in "unary key boxes" correspond exactly to inverse-functionality constraints on datatype properties. Such unrestricted key boxes make reasoning undecidable already if added to the basic description logic ALC, which is a fragment of OWL. However, the same paper points out restrictions to key statements under which reasoning becomes decidable, but if I'm not mistaken, there is no complete coverage of inverse-functional DPs in current reasoners.

> Punning in OWL-2 is interesting. We have the case (not in FOAF) where
> we have to cover two choices e.g. for classification scheme between an
> object property linking to a SKOS concept OR to a literal or string.
> Any link where I can find more information on this?

You could start here [3,4] and follow the references.

[1] http://www.w3.org/TR/2009/CR-owl2-syntax-20090611/

[2] C. Lutz, C. Areces, I. Horrocks, and U. Sattler. Keys, Nominals, and Concrete Domains. Journal of Artificial Intelligence Research, Vol. 23, pages 667-726, 2005. PDF: http://www.cs.man.ac.uk/~horrocks/Publications/download/2004/LAHS04a.pdf

[3] http://www.w3.org/TR/owl2-new-features/#F12:_Punning
[4] http://www.w3.org/2007/OWL/wiki/Punning

Cheers

Thomas

> Apologies for asking more...
>
> Thanks in advance
>
> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]
> ] On Behalf Of Thomas Schneider
> Sent: mardi, 21. septembre 2010 10:04
> To: Submit feedback for Protege 4.x
> Subject: Re: [p4-feedback] Use of inverseFunctional for data
> properties
>
>
> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>
>> Good morning.
>>
>> While looking at the FOAF ontology on Protégé 4.0, different
>> properties, although declared as data properties, would be displayed
>> as data and object properties.
>>
>> Would you confirm that this results from the fact that these
>> properties are also declared as inverseFunctional, which would seem
>> to make them look like object properties to Protégé?
>
> I don't know the FOAF ontology, but I suspect that the described
> behaviour is due to punning, which is allowed in OWL 2: two entities
> of different type (e.g. an object property and a data property) can
> have the same name. It is possible that, if the parser reads an
> InverseFunctional statement, that the property is punned and treated
> as an object property for this particular statement and as a data
> property otherwise.
>
> However, I can't say for sure that this is the actual reason.
>
>> It also seems, still on Protégé 4.0 that it is not proposed to
>> declare data properties as being inverseFunctional. Any reason?
>
> Neither in 4.0 nor in 4.1 can you declare data properties as inverse
> functional. The reason is that the OWL specification doesn't allow
> this. If it were allowed, reasoning would become an undecidable
> problem.
>
>> Interesting to note that reasoners accept all this.
>
> I'm not sure what exactly you mean. If you invoked the reasoner within
> Protégé, it worked on the representation of the ontology created by
> Protégé (or the underlying OWL API). Since the properties in question
> had been punned during parsing as described above, the reasoner got
> the InverseFunctional statements for object properties.
> However, if you invoked a reasoner directly, e.g., Pellet on the
> command line, then they might simply have used punning too.
>
> Cheers
>
> Thomas
>
>>
>> What do you think?
>>
>> Regards,
>>
>> Jean-pierre
>> -----------------------------------------
>> **************************************************
>> This email and any files transmitted with it are confidential and
>> intended solely for the use of the individual or entity to whom they
>> are addressed.
>> If you have received this email in error, please notify the system
>> manager.
>> This footnote also confirms that this email message has been swept by
>> the mailgateway
>> **************************************************
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
> +
> ----------------------------------------------------------------------+
> |  Dr Thomas Schneider                    schneider (at)  
> cs.man.ac.uk  |
> |  School of Computer Science       http://www.cs.man.ac.uk/ 
> ~schneidt  |
> |  Kilburn Building, Room 2.114                 phone +44 161  
> 2756136  |
> |  University of
> Manchester                                            |
> |  Oxford Road                                             _///
> _       |
> |  Manchester M13 9PL                                      
> (o~o)       |
> +-----------------------------------------------------oOOO--(_)--
> OOOo--+
>
> Fremantle (vb.)
>   To steal things not worth the bother of stealing. One steals cars,
>   money and silver. Book matches, airline eyepatches and individual
>   pots of Trust House Forte apricot jam are merely fremantled.
>
>                   Douglas Adams, John Lloyd: The Deeper Meaning of
> Liff
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback

+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff



_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Rinke Hoekstra-4
In reply to this post by Evain, Jean-Pierre
Hi Jean-Pierre,

On 21 sep 2010, at 10:20, Evain, Jean-Pierre wrote:
> Interesting to note that OWL (I guess inc. FULL) doesn't allow the use of inverseFunctional for data properties. I believe you ;-) but I need to check this myself to have the right arguments when explaining this. Does the spec say this specifically?

The structural specification [1] lists all allowed axioms on properties in sections 9.2 and 9.3. The list of allowed axioms for data properties is notably shorter and does not include inverseFunctional. However, the distinction between object properties and datatype properties does not hold in OWL 2 Full (as any RDF document can be interpreted as an OWL 2 Full document). So, feel free to use inverse functionality if you don't care about the OWL 2 DL restrictions.

Note, however, that Protege is an OWL 2 DL editor and imposes these restrictions for you (hence the punning).

> Punning in OWL-2 is interesting. We have the case (not in FOAF) where we have to cover two choices e.g. for classification scheme between an object property linking to a SKOS concept OR to a literal or string. Any link where I can find more information on this?

I would use punning for this, as it is the most elegant way to ensure interoperability between RDF-only tools and OWL 2 DL tools. You will have to take this duplicity into consideration when you define your class axioms, however.

-Rinke

[1] http://www.w3.org/TR/owl-syntax

>
> Apologies for asking more...
>
> Thanks in advance
>
> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf Of Thomas Schneider
> Sent: mardi, 21. septembre 2010 10:04
> To: Submit feedback for Protege 4.x
> Subject: Re: [p4-feedback] Use of inverseFunctional for data properties
>
>
> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>
>> Good morning.
>>
>> While looking at the FOAF ontology on Protégé 4.0, different
>> properties, although declared as data properties, would be displayed
>> as data and object properties.
>>
>> Would you confirm that this results from the fact that these
>> properties are also declared as inverseFunctional, which would seem to
>> make them look like object properties to Protégé?
>
> I don't know the FOAF ontology, but I suspect that the described behaviour is due to punning, which is allowed in OWL 2: two entities of different type (e.g. an object property and a data property) can have the same name. It is possible that, if the parser reads an InverseFunctional statement, that the property is punned and treated as an object property for this particular statement and as a data property otherwise.
>
> However, I can't say for sure that this is the actual reason.
>
>> It also seems, still on Protégé 4.0 that it is not proposed to declare
>> data properties as being inverseFunctional. Any reason?
>
> Neither in 4.0 nor in 4.1 can you declare data properties as inverse functional. The reason is that the OWL specification doesn't allow this. If it were allowed, reasoning would become an undecidable problem.
>
>> Interesting to note that reasoners accept all this.
>
> I'm not sure what exactly you mean. If you invoked the reasoner within Protégé, it worked on the representation of the ontology created by Protégé (or the underlying OWL API). Since the properties in question had been punned during parsing as described above, the reasoner got the InverseFunctional statements for object properties. However, if you invoked a reasoner directly, e.g., Pellet on the command line, then they might simply have used punning too.
>
> Cheers
>
> Thomas
>
>>
>> What do you think?
>>
>> Regards,
>>
>> Jean-pierre
>> -----------------------------------------
>> **************************************************
>> This email and any files transmitted with it are confidential and
>> intended solely for the use of the individual or entity to whom they
>> are addressed.
>> If you have received this email in error, please notify the system
>> manager.
>> This footnote also confirms that this email message has been swept by
>> the mailgateway
>> **************************************************
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
> +----------------------------------------------------------------------+
> |  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
> |  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
> |  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
> |  University of Manchester                                            |
> |  Oxford Road                                             _///_       |
> |  Manchester M13 9PL                                      (o~o)       |
> +-----------------------------------------------------oOOO--(_)--OOOo--+
>
> Fremantle (vb.)
>  To steal things not worth the bother of stealing. One steals cars,
>  money and silver. Book matches, airline eyepatches and individual
>  pots of Trust House Forte apricot jam are merely fremantled.
>
>                  Douglas Adams, John Lloyd: The Deeper Meaning of Liff
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback


---
Dr Rinke Hoekstra

AI Department         |   Leibniz Center for Law    
Faculty of Sciences   |   Faculty of Law            
Vrije Universiteit    |   Universiteit van Amsterdam
De Boelelaan 1081a    |   Kloveniersburgwal 48      
1081 HV Amsterdam     |   1012 CX  Amsterdam        
+31-(0)20-5987752     |   +31-(0)20-5253497        
[hidden email]    |   [hidden email]          

Homepage: http://www.few.vu.nl/~hoekstra





---
Dr Rinke Hoekstra

AI Department         |   Leibniz Center for Law    
Faculty of Sciences   |   Faculty of Law            
Vrije Universiteit    |   Universiteit van Amsterdam
De Boelelaan 1081a    |   Kloveniersburgwal 48      
1081 HV Amsterdam     |   1012 CX  Amsterdam        
+31-(0)20-5987752     |   +31-(0)20-5253497        
[hidden email]    |   [hidden email]          

Homepage: http://www.few.vu.nl/~hoekstra




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Evain, Jean-Pierre
Thanks Rinke.

This takes me rereading the reference documents. Sigh :-)

-----Original Message-----
From: [hidden email] [mailto:[hidden email]] On Behalf Of Rinke Hoekstra
Sent: mardi, 21. septembre 2010 10:57
To: Submit feedback for Protege 4.x
Subject: Re: [p4-feedback] Use of inverseFunctional for data properties

Hi Jean-Pierre,

On 21 sep 2010, at 10:20, Evain, Jean-Pierre wrote:
> Interesting to note that OWL (I guess inc. FULL) doesn't allow the use of inverseFunctional for data properties. I believe you ;-) but I need to check this myself to have the right arguments when explaining this. Does the spec say this specifically?

The structural specification [1] lists all allowed axioms on properties in sections 9.2 and 9.3. The list of allowed axioms for data properties is notably shorter and does not include inverseFunctional. However, the distinction between object properties and datatype properties does not hold in OWL 2 Full (as any RDF document can be interpreted as an OWL 2 Full document). So, feel free to use inverse functionality if you don't care about the OWL 2 DL restrictions.

Note, however, that Protege is an OWL 2 DL editor and imposes these restrictions for you (hence the punning).

> Punning in OWL-2 is interesting. We have the case (not in FOAF) where we have to cover two choices e.g. for classification scheme between an object property linking to a SKOS concept OR to a literal or string. Any link where I can find more information on this?

I would use punning for this, as it is the most elegant way to ensure interoperability between RDF-only tools and OWL 2 DL tools. You will have to take this duplicity into consideration when you define your class axioms, however.

-Rinke

[1] http://www.w3.org/TR/owl-syntax

>
> Apologies for asking more...
>
> Thanks in advance
>
> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf Of Thomas Schneider
> Sent: mardi, 21. septembre 2010 10:04
> To: Submit feedback for Protege 4.x
> Subject: Re: [p4-feedback] Use of inverseFunctional for data properties
>
>
> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>
>> Good morning.
>>
>> While looking at the FOAF ontology on Protégé 4.0, different
>> properties, although declared as data properties, would be displayed
>> as data and object properties.
>>
>> Would you confirm that this results from the fact that these
>> properties are also declared as inverseFunctional, which would seem to
>> make them look like object properties to Protégé?
>
> I don't know the FOAF ontology, but I suspect that the described behaviour is due to punning, which is allowed in OWL 2: two entities of different type (e.g. an object property and a data property) can have the same name. It is possible that, if the parser reads an InverseFunctional statement, that the property is punned and treated as an object property for this particular statement and as a data property otherwise.
>
> However, I can't say for sure that this is the actual reason.
>
>> It also seems, still on Protégé 4.0 that it is not proposed to declare
>> data properties as being inverseFunctional. Any reason?
>
> Neither in 4.0 nor in 4.1 can you declare data properties as inverse functional. The reason is that the OWL specification doesn't allow this. If it were allowed, reasoning would become an undecidable problem.
>
>> Interesting to note that reasoners accept all this.
>
> I'm not sure what exactly you mean. If you invoked the reasoner within Protégé, it worked on the representation of the ontology created by Protégé (or the underlying OWL API). Since the properties in question had been punned during parsing as described above, the reasoner got the InverseFunctional statements for object properties. However, if you invoked a reasoner directly, e.g., Pellet on the command line, then they might simply have used punning too.
>
> Cheers
>
> Thomas
>
>>
>> What do you think?
>>
>> Regards,
>>
>> Jean-pierre
>> -----------------------------------------
>> **************************************************
>> This email and any files transmitted with it are confidential and
>> intended solely for the use of the individual or entity to whom they
>> are addressed.
>> If you have received this email in error, please notify the system
>> manager.
>> This footnote also confirms that this email message has been swept by
>> the mailgateway
>> **************************************************
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
> +----------------------------------------------------------------------+
> |  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
> |  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
> |  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
> |  University of Manchester                                            |
> |  Oxford Road                                             _///_       |
> |  Manchester M13 9PL                                      (o~o)       |
> +-----------------------------------------------------oOOO--(_)--OOOo--+
>
> Fremantle (vb.)
>  To steal things not worth the bother of stealing. One steals cars,
>  money and silver. Book matches, airline eyepatches and individual
>  pots of Trust House Forte apricot jam are merely fremantled.
>
>                  Douglas Adams, John Lloyd: The Deeper Meaning of Liff
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback


---
Dr Rinke Hoekstra

AI Department         |   Leibniz Center for Law    
Faculty of Sciences   |   Faculty of Law            
Vrije Universiteit    |   Universiteit van Amsterdam
De Boelelaan 1081a    |   Kloveniersburgwal 48      
1081 HV Amsterdam     |   1012 CX  Amsterdam        
+31-(0)20-5987752     |   +31-(0)20-5253497        
[hidden email]    |   [hidden email]          

Homepage: http://www.few.vu.nl/~hoekstra





---
Dr Rinke Hoekstra

AI Department         |   Leibniz Center for Law    
Faculty of Sciences   |   Faculty of Law            
Vrije Universiteit    |   Universiteit van Amsterdam
De Boelelaan 1081a    |   Kloveniersburgwal 48      
1081 HV Amsterdam     |   1012 CX  Amsterdam        
+31-(0)20-5987752     |   +31-(0)20-5253497        
[hidden email]    |   [hidden email]          

Homepage: http://www.few.vu.nl/~hoekstra




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Rinke Hoekstra-4
In reply to this post by Evain, Jean-Pierre
Hi,

Note that OWL 2 DL does allow for 'easy keys', a restricted form of inverse functionality [1] that is defined as a class axiom (i.e. a property functions as an inverse functional property only for members of that class). If you define a key on owl:Thing the 'functionality' is almost identical to that of a real inverse functional property. I believe Protege 4.1 has support for this.

Easy keys have a different syntax in the RDF serialization, though, so you'll have to check that [2] to see whether it suits your needs.

-Rinke

[1] http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/#Keys 
[2] http://www.w3.org/TR/2009/REC-owl2-mapping-to-rdf-20091027/#Translation_of_Axioms_without_Annotations



On 21 sep 2010, at 10:53, Evain, Jean-Pierre wrote:

> Appreciate it, thanks, JP
>
> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]] On Behalf Of Thomas Schneider
> Sent: mardi, 21. septembre 2010 10:47
> To: Submit feedback for Protege 4.x
> Subject: Re: [p4-feedback] Use of inverseFunctional for data properties
>
> Hi Jean-Pierre,
>
> On 21 Sep 2010, at 09:20, Evain, Jean-Pierre wrote:
>
>> Interesting to note that OWL (I guess inc. FULL) doesn't allow the use
>> of inverseFunctional for data properties. I believe you ;-) but I need
>> to check this myself to have the right arguments when explaining this.
>> Does the spec say this specifically?
>
> I haven't found a note on this in the structural specification document [1]. However, my claim was based on the results in the technical paper [2], where the expressions allowed in "unary key boxes" correspond exactly to inverse-functionality constraints on datatype properties. Such unrestricted key boxes make reasoning undecidable already if added to the basic description logic ALC, which is a fragment of OWL. However, the same paper points out restrictions to key statements under which reasoning becomes decidable, but if I'm not mistaken, there is no complete coverage of inverse-functional DPs in current reasoners.
>
>> Punning in OWL-2 is interesting. We have the case (not in FOAF) where
>> we have to cover two choices e.g. for classification scheme between an
>> object property linking to a SKOS concept OR to a literal or string.
>> Any link where I can find more information on this?
>
> You could start here [3,4] and follow the references.
>
> [1] http://www.w3.org/TR/2009/CR-owl2-syntax-20090611/
>
> [2] C. Lutz, C. Areces, I. Horrocks, and U. Sattler. Keys, Nominals, and Concrete Domains. Journal of Artificial Intelligence Research, Vol. 23, pages 667-726, 2005. PDF: http://www.cs.man.ac.uk/~horrocks/Publications/download/2004/LAHS04a.pdf
>
> [3] http://www.w3.org/TR/owl2-new-features/#F12:_Punning
> [4] http://www.w3.org/2007/OWL/wiki/Punning
>
> Cheers
>
> Thomas
>
>> Apologies for asking more...
>>
>> Thanks in advance
>>
>> -----Original Message-----
>> From: [hidden email]
>> [mailto:[hidden email]
>> ] On Behalf Of Thomas Schneider
>> Sent: mardi, 21. septembre 2010 10:04
>> To: Submit feedback for Protege 4.x
>> Subject: Re: [p4-feedback] Use of inverseFunctional for data
>> properties
>>
>>
>> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>>
>>> Good morning.
>>>
>>> While looking at the FOAF ontology on Protégé 4.0, different
>>> properties, although declared as data properties, would be displayed
>>> as data and object properties.
>>>
>>> Would you confirm that this results from the fact that these
>>> properties are also declared as inverseFunctional, which would seem
>>> to make them look like object properties to Protégé?
>>
>> I don't know the FOAF ontology, but I suspect that the described
>> behaviour is due to punning, which is allowed in OWL 2: two entities
>> of different type (e.g. an object property and a data property) can
>> have the same name. It is possible that, if the parser reads an
>> InverseFunctional statement, that the property is punned and treated
>> as an object property for this particular statement and as a data
>> property otherwise.
>>
>> However, I can't say for sure that this is the actual reason.
>>
>>> It also seems, still on Protégé 4.0 that it is not proposed to
>>> declare data properties as being inverseFunctional. Any reason?
>>
>> Neither in 4.0 nor in 4.1 can you declare data properties as inverse
>> functional. The reason is that the OWL specification doesn't allow
>> this. If it were allowed, reasoning would become an undecidable
>> problem.
>>
>>> Interesting to note that reasoners accept all this.
>>
>> I'm not sure what exactly you mean. If you invoked the reasoner within
>> Protégé, it worked on the representation of the ontology created by
>> Protégé (or the underlying OWL API). Since the properties in question
>> had been punned during parsing as described above, the reasoner got
>> the InverseFunctional statements for object properties.
>> However, if you invoked a reasoner directly, e.g., Pellet on the
>> command line, then they might simply have used punning too.
>>
>> Cheers
>>
>> Thomas
>>
>>>
>>> What do you think?
>>>
>>> Regards,
>>>
>>> Jean-pierre
>>> -----------------------------------------
>>> **************************************************
>>> This email and any files transmitted with it are confidential and
>>> intended solely for the use of the individual or entity to whom they
>>> are addressed.
>>> If you have received this email in error, please notify the system
>>> manager.
>>> This footnote also confirms that this email message has been swept by
>>> the mailgateway
>>> **************************************************
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>
>> +
>> ----------------------------------------------------------------------+
>> |  Dr Thomas Schneider                    schneider (at)  
>> cs.man.ac.uk  |
>> |  School of Computer Science       http://www.cs.man.ac.uk/ 
>> ~schneidt  |
>> |  Kilburn Building, Room 2.114                 phone +44 161  
>> 2756136  |
>> |  University of
>> Manchester                                            |
>> |  Oxford Road                                             _///
>> _       |
>> |  Manchester M13 9PL                                      
>> (o~o)       |
>> +-----------------------------------------------------oOOO--(_)--
>> OOOo--+
>>
>> Fremantle (vb.)
>>  To steal things not worth the bother of stealing. One steals cars,
>>  money and silver. Book matches, airline eyepatches and individual
>>  pots of Trust House Forte apricot jam are merely fremantled.
>>
>>                  Douglas Adams, John Lloyd: The Deeper Meaning of
>> Liff
>>
>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
> +----------------------------------------------------------------------+
> |  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
> |  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
> |  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
> |  University of Manchester                                            |
> |  Oxford Road                                             _///_       |
> |  Manchester M13 9PL                                      (o~o)       |
> +-----------------------------------------------------oOOO--(_)--OOOo--+
>
> Fremantle (vb.)
>   To steal things not worth the bother of stealing. One steals cars,
>   money and silver. Book matches, airline eyepatches and individual
>   pots of Trust House Forte apricot jam are merely fremantled.
>
>                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback


---
Dr Rinke Hoekstra

AI Department         |   Leibniz Center for Law    
Faculty of Sciences   |   Faculty of Law            
Vrije Universiteit    |   Universiteit van Amsterdam
De Boelelaan 1081a    |   Kloveniersburgwal 48      
1081 HV Amsterdam     |   1012 CX  Amsterdam        
+31-(0)20-5987752     |   +31-(0)20-5253497        
[hidden email]    |   [hidden email]          

Homepage: http://www.few.vu.nl/~hoekstra




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Property union

Thomas Schneider-5
In reply to this post by regmito
Hi Cristian, Igor,

On 20 Sep 2010, at 17:38, Cristian Cocos wrote:

> Here's an OWL 2 question: can I say that a property is the union of  
> two (or
> more) properties ? (I am, in effect, not sure if SROIQ can  
> accommodate this,
> can it?)


Another option, which slightly extends the suggestions given so far,  
would be to look at the usage of such property unions.

* If they are only used in existential statements, one could replace  
such a statement with a class union:
   "(p or q) some C"  is equivalent to  "p some C or q some C".

* If they are only used in universal statements, one could replace  
such a statement with a class intersection:
   "(p or q) only C"  is equivalent to  "p only C and q only C".

These replacements become combinatorially more complex if cardinality  
restrictions are involved, though.

On 21 Sep 2010, at 02:57, regmito wrote:

> For more systematic approach, property union and intersection should  
> be allowed as property expressions that are similar to class  
> expressions: e.g. (a or b); (a and b). However the OWL 2  
> specification misses this.

I don't think it's that easy to add property intersection and union to  
OWL 2. For these features to be harmless, one would at least need to  
restrict the interactions with cardinality restrictions, i.e., extend  
the concept of simple properties to union/intersection.

However, this doesn't seem to suffice for property intersection. If I  
didn't make a mistake, then, even without number restrictions, it's  
possible to express tilings and therefore to cross the undecidability  
border. I'm currently not sure if this can also happen with property  
union ...

Cheers

Thomas

>
> Igor
>
>
> Dr Igor Toujilov
> Senior Research Fellow
> Centre for Health Informatics and Multiprofessional Education
> University College London
> Highgate Hill
> London, UK
> N19 5LW
>
> Tel: +44 (0)20 7288 3960
>
>
> ________________________________________
> From: [hidden email] [[hidden email]
> ] on behalf of Alan Ruttenberg [[hidden email]]
> Sent: 20 September 2010 20:11
> To: Submit feedback for Protege 4.x
> Cc: [hidden email]
> Subject: Re: [p4-feedback] Property union
>
> If :a and :b are the properties that you want to union, then an
> approximation is to make the union :u the superproperty of both. This
> will break down if you either make property assertions using :u
> directly or have axioms that might entail such assertions.
>
> -Alan
>
> On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>
> wrote:
>
>>
>> I believe that the answer is that you can't do this.  While there
>> are several experts here who might be  able to either correct me or
>> give a stronger answer, I suspect that a better list for this type
>> of question is the owl-dev mailing list [1].
>>
>> -Timothy
>>
>> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>>
>>
>> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>>> Here's an OWL 2 question: can I say that a property is the union of
>>> two (or
>>> more) properties ? (I am, in effect, not sure if SROIQ can
>>> accommodate this,
>>> can it?)
>>>
>>> C
>>>
>>> Cristian Cocos
>>> Post Doctoral Fellow
>>> Centre for Logic and Information, StFX University 54B St. Mary's
>>> Street,
>>> Antigonish NS, Canada B2G 2W5
>>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>>
>>> Current research: "Building Decision-Support Through Dynamic  
>>> Workflow
>>> Systems for Health Care"
>>>
>>>
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

PGP.sig (210 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Property union

regmito
Hi All,
Thomas, it is interesting what you said, and it would be nice to look at this in the context of my current proposal (see below). Now I realise how naïve my efforts were to implement the property union so straightforwardly.

Alan, Thomas, thank you for the counterexamples that helped me to understand we need tracking both ends of the property arc with variables. So, it seems impossible to implement this without SWRL.

Then I have come to the idea of meta-modelling that is presented below. The meta-model is:

ObjectProperty: has_subject
    Characteristics: Functional

ObjectProperty: has_value
    Characteristics: Functional

Class: property
    EquivalentTo: (has_subject some owl:Thing) and (has_value some owl:Thing)

Class: property_a
Class: property_b

Class: property_u
    EquivalentTo: property_a or property_b

property
    property_u
        property_a
        property_b

This meta-model can be generated automatically, based on the template.
Then we need an object property interface to interact conveniently with the meta-model. This interface can be also generated automatically:

ObjectProperty: u
ObjectProperty: a
ObjectProperty: b

u
      a
      b

property_a(?p), has_subject(?p, ?s), has_value(?p, ?v) -> a(?s, ?v).
property_b(?p), has_subject(?p, ?s), has_value(?p, ?v) -> b(?s, ?v).
property_u(?p), has_subject(?p, ?s), has_value(?p, ?v) -> u(?s, ?v).

The properties should be instantiated in the meta-model, not in the interface. A joint OWL file, containing the meta-model, interface, and instances, is attached.

Cheers,
Igor
________________________________________
From: [hidden email] [[hidden email]] on behalf of Thomas Schneider [[hidden email]]
Sent: 21 September 2010 16:36
To: Submit feedback for Protege 4.x
Subject: Re: [p4-feedback] Property union

Hi Cristian, Igor,

On 20 Sep 2010, at 17:38, Cristian Cocos wrote:

> Here's an OWL 2 question: can I say that a property is the union of
> two (or
> more) properties ? (I am, in effect, not sure if SROIQ can
> accommodate this,
> can it?)


Another option, which slightly extends the suggestions given so far,
would be to look at the usage of such property unions.

* If they are only used in existential statements, one could replace
such a statement with a class union:
   "(p or q) some C"  is equivalent to  "p some C or q some C".

* If they are only used in universal statements, one could replace
such a statement with a class intersection:
   "(p or q) only C"  is equivalent to  "p only C and q only C".

These replacements become combinatorially more complex if cardinality
restrictions are involved, though.

On 21 Sep 2010, at 02:57, regmito wrote:

> For more systematic approach, property union and intersection should
> be allowed as property expressions that are similar to class
> expressions: e.g. (a or b); (a and b). However the OWL 2
> specification misses this.

I don't think it's that easy to add property intersection and union to
OWL 2. For these features to be harmless, one would at least need to
restrict the interactions with cardinality restrictions, i.e., extend
the concept of simple properties to union/intersection.

However, this doesn't seem to suffice for property intersection. If I
didn't make a mistake, then, even without number restrictions, it's
possible to express tilings and therefore to cross the undecidability
border. I'm currently not sure if this can also happen with property
union ...

Cheers

Thomas

>
> Igor
>
>
> Dr Igor Toujilov
> Senior Research Fellow
> Centre for Health Informatics and Multiprofessional Education
> University College London
> Highgate Hill
> London, UK
> N19 5LW
>
> Tel: +44 (0)20 7288 3960
>
>
> ________________________________________
> From: [hidden email] [[hidden email]
> ] on behalf of Alan Ruttenberg [[hidden email]]
> Sent: 20 September 2010 20:11
> To: Submit feedback for Protege 4.x
> Cc: [hidden email]
> Subject: Re: [p4-feedback] Property union
>
> If :a and :b are the properties that you want to union, then an
> approximation is to make the union :u the superproperty of both. This
> will break down if you either make property assertions using :u
> directly or have axioms that might entail such assertions.
>
> -Alan
>
> On Sep 20, 2010, at 2:31 PM, Timothy Redmond <[hidden email]>
> wrote:
>
>>
>> I believe that the answer is that you can't do this.  While there
>> are several experts here who might be  able to either correct me or
>> give a stronger answer, I suspect that a better list for this type
>> of question is the owl-dev mailing list [1].
>>
>> -Timothy
>>
>> [1] http://lists.w3.org/Archives/Public/public-owl-dev/
>>
>>
>> On 09/20/2010 09:38 AM, Cristian Cocos wrote:
>>> Here's an OWL 2 question: can I say that a property is the union of
>>> two (or
>>> more) properties ? (I am, in effect, not sure if SROIQ can
>>> accommodate this,
>>> can it?)
>>>
>>> C
>>>
>>> Cristian Cocos
>>> Post Doctoral Fellow
>>> Centre for Logic and Information, StFX University 54B St. Mary's
>>> Street,
>>> Antigonish NS, Canada B2G 2W5
>>> Tel: + 1 (902) 867-4931, Fax: +1 (902) 867-1397
>>>
>>> Current research: "Building Decision-Support Through Dynamic
>>> Workflow
>>> Systems for Health Care"
>>>
>>>
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

property_union.owl (19K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Alan Ruttenberg-2
In reply to this post by Thomas Schneider-5
On Tue, Sep 21, 2010 at 4:04 AM, Thomas Schneider <[hidden email]> wrote:

>
> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>
>> Good morning.
>>
>> While looking at the FOAF ontology on Protégé 4.0, different properties,
>> although declared as data properties, would be displayed as data and object
>> properties.
>>
>> Would you confirm that this results from the fact that these properties
>> are also declared as inverseFunctional, which would seem to make them look
>> like object properties to Protégé?
>
> I don't know the FOAF ontology, but I suspect that the described behaviour
> is due to punning, which is allowed in OWL 2: two entities of different type
> (e.g. an object property and a data property) can have the same name.

Object and Data properties can not be punned.
http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/#Typing_Constraints_of_OWL_2_DL

> It is
> possible that, if the parser reads an InverseFunctional statement, that the
> property is punned and treated as an object property for this particular
> statement and as a data property otherwise.
>
> However, I can't say for sure that this is the actual reason.
>
>> It also seems, still on Protégé 4.0 that it is not proposed to declare
>> data properties as being inverseFunctional. Any reason?
>
> Neither in 4.0 nor in 4.1 can you declare data properties as inverse
> functional. The reason is that the OWL specification doesn't allow this. If
> it were allowed, reasoning would become an undecidable problem.
>
>> Interesting to note that reasoners accept all this.
>
> I'm not sure what exactly you mean. If you invoked the reasoner within
> Protégé, it worked on the representation of the ontology created by Protégé
> (or the underlying OWL API). Since the properties in question had been
> punned during parsing as described above, the reasoner got the
> InverseFunctional statements for object properties. However, if you invoked
> a reasoner directly, e.g., Pellet on the command line, then they might
> simply have used punning too.
>
> Cheers
>
> Thomas
>
>>
>> What do you think?
>>
>> Regards,
>>
>> Jean-pierre
>> -----------------------------------------
>> **************************************************
>> This email and any files transmitted with it
>> are confidential and intended solely for the
>> use of the individual or entity to whom they
>> are addressed.
>> If you have received this email in error,
>> please notify the system manager.
>> This footnote also confirms that this email
>> message has been swept by the mailgateway
>> **************************************************
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
> +----------------------------------------------------------------------+
> |  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
> |  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
> |  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
> |  University of Manchester                                            |
> |  Oxford Road                                             _///_       |
> |  Manchester M13 9PL                                      (o~o)       |
> +-----------------------------------------------------oOOO--(_)--OOOo--+
>
> Fremantle (vb.)
>  To steal things not worth the bother of stealing. One steals cars,
>  money and silver. Book matches, airline eyepatches and individual
>  pots of Trust House Forte apricot jam are merely fremantled.
>
>                  Douglas Adams, John Lloyd: The Deeper Meaning of Liff
>
>
>
>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
>
_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Thomas Schneider-5

On 22 Sep 2010, at 03:52, Alan Ruttenberg wrote:

> On Tue, Sep 21, 2010 at 4:04 AM, Thomas Schneider <[hidden email]
> > wrote:
>>
>> On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:
>>
>>> Good morning.
>>>
>>> While looking at the FOAF ontology on Protégé 4.0, different  
>>> properties,
>>> although declared as data properties, would be displayed as data  
>>> and object
>>> properties.
>>>
>>> Would you confirm that this results from the fact that these  
>>> properties
>>> are also declared as inverseFunctional, which would seem to make  
>>> them look
>>> like object properties to Protégé?
>>
>> I don't know the FOAF ontology, but I suspect that the described  
>> behaviour
>> is due to punning, which is allowed in OWL 2: two entities of  
>> different type
>> (e.g. an object property and a data property) can have the same name.
>
> Object and Data properties can not be punned.
> http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/#Typing_Constraints_of_OWL_2_DL
Hmmm, sorry for overlooking this ...

Then could the described behaviour be a parsing problem? It might be  
worth trying to open the same ontology with the recent Protégé version  
(4.1 beta), where conformance with the OWL specification has been  
improved.

But maybe someone from the Protégé team should comment on this.

Cheers

Thomas

>
>> It is
>> possible that, if the parser reads an InverseFunctional statement,  
>> that the
>> property is punned and treated as an object property for this  
>> particular
>> statement and as a data property otherwise.
>>
>> However, I can't say for sure that this is the actual reason.
>>
>>> It also seems, still on Protégé 4.0 that it is not proposed to  
>>> declare
>>> data properties as being inverseFunctional. Any reason?
>>
>> Neither in 4.0 nor in 4.1 can you declare data properties as inverse
>> functional. The reason is that the OWL specification doesn't allow  
>> this. If
>> it were allowed, reasoning would become an undecidable problem.
>>
>>> Interesting to note that reasoners accept all this.
>>
>> I'm not sure what exactly you mean. If you invoked the reasoner  
>> within
>> Protégé, it worked on the representation of the ontology created by  
>> Protégé
>> (or the underlying OWL API). Since the properties in question had  
>> been
>> punned during parsing as described above, the reasoner got the
>> InverseFunctional statements for object properties. However, if you  
>> invoked
>> a reasoner directly, e.g., Pellet on the command line, then they  
>> might
>> simply have used punning too.
>>
>> Cheers
>>
>> Thomas
>>
>>>
>>> What do you think?
>>>
>>> Regards,
>>>
>>> Jean-pierre
>>> -----------------------------------------
>>> **************************************************
>>> This email and any files transmitted with it
>>> are confidential and intended solely for the
>>> use of the individual or entity to whom they
>>> are addressed.
>>> If you have received this email in error,
>>> please notify the system manager.
>>> This footnote also confirms that this email
>>> message has been swept by the mailgateway
>>> **************************************************
>>>
>>> _______________________________________________
>>> p4-feedback mailing list
>>> [hidden email]
>>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>
>> +
>> ----------------------------------------------------------------------+
>> |  Dr Thomas Schneider                    schneider (at)  
>> cs.man.ac.uk  |
>> |  School of Computer Science       http://www.cs.man.ac.uk/ 
>> ~schneidt  |
>> |  Kilburn Building, Room 2.114                 phone +44 161  
>> 2756136  |
>> |  University of  
>> Manchester                                            |
>> |  Oxford Road                                             _///
>> _       |
>> |  Manchester M13 9PL                                      
>> (o~o)       |
>> +-----------------------------------------------------oOOO--(_)--
>> OOOo--+
>>
>> Fremantle (vb.)
>>  To steal things not worth the bother of stealing. One steals cars,
>>  money and silver. Book matches, airline eyepatches and individual
>>  pots of Trust House Forte apricot jam are merely fremantled.
>>
>>                  Douglas Adams, John Lloyd: The Deeper Meaning of  
>> Liff
>>
>>
>>
>>
>> _______________________________________________
>> p4-feedback mailing list
>> [hidden email]
>> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>>
>>
> _______________________________________________
> p4-feedback mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback
+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
   To steal things not worth the bother of stealing. One steals cars,
   money and silver. Book matches, airline eyepatches and individual
   pots of Trust House Forte apricot jam are merely fremantled.

                   Douglas Adams, John Lloyd: The Deeper Meaning of Liff




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

PGP.sig (210 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Use of inverseFunctional for data properties

Timothy Redmond

Then could the described behaviour be a parsing problem? It might be worth trying to open the same ontology with the recent Protégé version (4.1 beta), where conformance with the OWL specification has been improved.

But maybe someone from the Protégé team should comment on this.

It looks like the latest version of the OWL api allows punning between object properties and data properties.  But it also appears that it gets pretty confused in this situation.

-Timothy



On 09/22/2010 12:33 AM, Thomas Schneider wrote:

On 22 Sep 2010, at 03:52, Alan Ruttenberg wrote:

On Tue, Sep 21, 2010 at 4:04 AM, Thomas Schneider [hidden email] wrote:

On 21 Sep 2010, at 08:43, Evain, Jean-Pierre wrote:

Good morning.

While looking at the FOAF ontology on Protégé 4.0, different properties,
although declared as data properties, would be displayed as data and object
properties.

Would you confirm that this results from the fact that these properties
are also declared as inverseFunctional, which would seem to make them look
like object properties to Protégé?

I don't know the FOAF ontology, but I suspect that the described behaviour
is due to punning, which is allowed in OWL 2: two entities of different type
(e.g. an object property and a data property) can have the same name.

Object and Data properties can not be punned.
http://www.w3.org/TR/2009/REC-owl2-syntax-20091027/#Typing_Constraints_of_OWL_2_DL

Hmmm, sorry for overlooking this ...

Then could the described behaviour be a parsing problem? It might be worth trying to open the same ontology with the recent Protégé version (4.1 beta), where conformance with the OWL specification has been improved.

But maybe someone from the Protégé team should comment on this.

Cheers

Thomas


It is
possible that, if the parser reads an InverseFunctional statement, that the
property is punned and treated as an object property for this particular
statement and as a data property otherwise.

However, I can't say for sure that this is the actual reason.

It also seems, still on Protégé 4.0 that it is not proposed to declare
data properties as being inverseFunctional. Any reason?

Neither in 4.0 nor in 4.1 can you declare data properties as inverse
functional. The reason is that the OWL specification doesn't allow this. If
it were allowed, reasoning would become an undecidable problem.

Interesting to note that reasoners accept all this.

I'm not sure what exactly you mean. If you invoked the reasoner within
Protégé, it worked on the representation of the ontology created by Protégé
(or the underlying OWL API). Since the properties in question had been
punned during parsing as described above, the reasoner got the
InverseFunctional statements for object properties. However, if you invoked
a reasoner directly, e.g., Pellet on the command line, then they might
simply have used punning too.

Cheers

Thomas


What do you think?

Regards,

Jean-pierre
-----------------------------------------
**************************************************
This email and any files transmitted with it
are confidential and intended solely for the
use of the individual or entity to whom they
are addressed.
If you have received this email in error,
please notify the system manager.
This footnote also confirms that this email
message has been swept by the mailgateway
**************************************************

_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
 To steal things not worth the bother of stealing. One steals cars,
 money and silver. Book matches, airline eyepatches and individual
 pots of Trust House Forte apricot jam are merely fremantled.

                 Douglas Adams, John Lloyd: The Deeper Meaning of Liff




_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback


_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback

+----------------------------------------------------------------------+
|  Dr Thomas Schneider                    schneider (at) cs.man.ac.uk  |
|  School of Computer Science       http://www.cs.man.ac.uk/~schneidt  |
|  Kilburn Building, Room 2.114                 phone +44 161 2756136  |
|  University of Manchester                                            |
|  Oxford Road                                             _///_       |
|  Manchester M13 9PL                                      (o~o)       |
+-----------------------------------------------------oOOO--(_)--OOOo--+

Fremantle (vb.)
  To steal things not worth the bother of stealing. One steals cars,
  money and silver. Book matches, airline eyepatches and individual
  pots of Trust House Forte apricot jam are merely fremantled.

                  Douglas Adams, John Lloyd: The Deeper Meaning of Liff



_______________________________________________ p4-feedback mailing list [hidden email] https://mailman.stanford.edu/mailman/listinfo/p4-feedback


_______________________________________________
p4-feedback mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/p4-feedback
12