Re: [protege-owl] Protege 4 - Instance editor

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

Re: [protege-owl] Protege 4 - Instance editor

Bo De Lange-2
Hi Jesus and Nick (also picking up on a post in the P4 list by Mathe Maema 2008/10/13),


I was enquiring about  more or less the same feature when I was in Manchester  in November 2007 and I would still very much like to see something of this kind realised in P4.

Instance forms creation wrt an ontology is a non-trivial exercise as OWL does not act in a constraints manner as many people expect.

We have ported our ontology from Protege Frames and may well be misusing the contstraints that were possible there as domains and ranges set for properties in OWL.

Let's recap on individuals, classes and properties in OWL  (quoting from the OWL tutorial and course info).

  1. Classes are sets of individuals
  2. Properties are binary relations on individuals
  3. Classes may be built up of descriptions that specify the conditions that must be met by an individual for it to be a member of the class. These descriptions can be property restictions 
  4. Properties may have a domain and a range specified. Properties then link individuals from the domain class(es) to individuals from the range class(es).

Domains and ranges are NOT constraints to be checked (as they are in Protege Frames). They are rather used as axioms in reasoning. E.g. in the Pizza ontology if the property hasTopping has the domain set as Pizza (with range PizzaTopping) and the hasTopping property was then applied to (used to restrict) the Class IceCream, this would NOT normally generate an error by the reasoner, but rather infer that IceCream must be a subclass of Pizza! An error would be generated by the reasoner when the classes Pizza and IceCream had been made disjoint.

In other words, when domains and ranges are set for a property, and this property is used in a class restriction or definition,  the indivduals that are used on the left hand side in this restriction are inferred to be members of the Domain class and the individuals on the right hand side are inferred to be members of the Range class, regardless of what classes you are applying the property to.

This is why it is generally discouraged to set domains and ranges for object properties.


However, our ontology does -deliberately- use domains and ranges to facilitate individual input in P3.4. None of the properties with domain and range set is used in restricting classes (other than the domain and range classes). In fact, we have transformed most of our object and data type properties to annotation type properties, since we did not plan to use these properties for automated classification by the reasoner, and wanted to ease the burden on the reasoner in our sizeable ontology. (Which reminds me: why does the P4 OWL tutorial -edition 1.1- say on P74 that annotation properties must not have a domain and a range set for them?).

So in our case: yes, the type of data the expert is entering is regular in that created individuals are of a single class. I do not see myself developing the view plugin, so I'm hoping others with more experience in plugin development will pick this issue up in the near future.

Bo

PS I've cross-posted this to the P4 feedback list where I think it belongs.





On Fri, Nov 7, 2008 at 3:21 PM, Nick Drummond <[hidden email]> wrote:
Hi Jesus,

Protege 4.0 has not been built with instance input in mind. Hence, there is currently no forms editor in P4.
Instance forms creation wrt an ontology is a non-trivial exercise as OWL does not act in a constraints manner as many people expect.
The forms mechanism in P3 is the result of many years of code development, but even it does not map 100% to creating OWL individuals.
If the type of data your expert is entering is regular (ie they are always creating instances of a sincle class) then coding a custom form as a view plugin would be straightforward.

Nick

2008/11/7 Jesus Bisbal <[hidden email]>
Dear all,
   I´m sure this must be a very basic question but with a quick look at
the mailing list, I could not find a tread on this.

   I have created a not very complex ontology in OWL, with Protégé 4.0
beta. Now it must be populated, which will take a significant amount of
time.
   Populating should be done by a domain expert, not with particular
technical skills. I want to make it simple for him. So, the ideal way
would be for example the "Forms" found in Protégé 3.4 and 3.1. Is there
no equivalent in Protégé 4.0? Is it under development or I should do it
in a completely different way
   Asking a non technical user to populate the ontology thinking about
properties, for example, does not seem appropriate.

   Many thanks,

Jesús

--
______________________________________________________________________
Jesús Bisbal-Riera                   http://www.dtic.upf.edu/~jbisbal
Departament de Tecnologies de la Informació i les Comunicacions
Universitat Pompeu Fabra             http://www.upf.edu
Passeig de Circumval·lació, 8
08003 Barcelona                      Work Ph: +34 93 542 29 51 / 25 00
Spain                                Fax:     +34 93 542 25 17

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

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03


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

Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03



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

Re: [protege-owl] Protege 4 - Instance editor

Bijan Parsia-3
Hi Bo,

I'm sorry I don't have time to sort through your issues in detail.

Using owl axioms to inform knowledge acquisition is a time honored  
practice. It does require, at the moment, an "extra owl" step. Alan  
Rector calls this "sanctioning".

There is support coming in OWL 2 for at least a syntactic distinction  
between sanctioning axioms and other sorts:
        http://www.w3.org/2007/OWL/wiki/Syntax#Annotation_Property_Axioms

Furthermore, there is some move toward allowing epistemicish  
constraints in as an extension of OWL, e.g.:
        http://www.webont.org/owled/2008/papers/owled2008eu_submission_30.pdf

And we have a sorta moribund (but highly successful) task force:
        http://www.webont.org/owled/2008/papers/owled2008eu_submission_30.pdf

Hope this helps.

Cheers,
Bijan.

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

Re: [protege-owl] Protege 4 - Instance editor

Bo De Lange-2
Thanks for your insights and the links, Bijan. The link you gave for the taskforce was the same as for the integrity constraints (i.e. the OWLED2008 Sirin pdf). Could you please repost the proper link?

Thanks,

Bo

On Wed, Nov 12, 2008 at 2:38 PM, Bijan Parsia <[hidden email]> wrote:
Hi Bo,

I'm sorry I don't have time to sort through your issues in detail.

Using owl axioms to inform knowledge acquisition is a time honored
practice. It does require, at the moment, an "extra owl" step. Alan
Rector calls this "sanctioning".

There is support coming in OWL 2 for at least a syntactic distinction
between sanctioning axioms and other sorts:
       http://www.w3.org/2007/OWL/wiki/Syntax#Annotation_Property_Axioms

Furthermore, there is some move toward allowing epistemicish
constraints in as an extension of OWL, e.g.:
       http://www.webont.org/owled/2008/papers/owled2008eu_submission_30.pdf

And we have a sorta moribund (but highly successful) task force:
       http://www.webont.org/owled/2008/papers/owled2008eu_submission_30.pdf

Hope this helps.

Cheers,
Bijan.

_______________________________________________
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: [protege-owl] Protege 4 - Instance editor

Bijan Parsia-3
On 12 Nov 2008, at 14:10, Bo De Lange wrote:

> Thanks for your insights and the links, Bijan. The link you gave  
> for the taskforce was the same as for the integrity constraints  
> (i.e. the OWLED2008 Sirin pdf). Could you please repost the proper  
> link?

Oops:
        http://wiki.webont.org/page/DatabasEsque
Cheers,
Bijan.

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