Quantcast

Questions about Data/Object Property Inferences between 2 Ontologies

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Questions about Data/Object Property Inferences between 2 Ontologies

sizedam
This post was updated on .
I am preparing to reconcile 2 different vocabularies within the same domain, but from 2 different organizations. I put together a little experiment to help decide whether to merge the vocabularies into 1 ontology or keep them separate and compare them via import. It is mostly working as I expected, but I have some questions.

Creatures Ontology:
* Class BodySurface with individuals furry & scaly
* Object Property hasBodySurface
* Class Creature with individuals squirrel & dog, each hasBodySurface furry

Animals Ontology:
* Data Property hasGlossaryDefinition
* Class Animal with sub-class Mammal, with individuals squirrel & canine, each with appropriate GlossaryDefinition values

Observations:
1. I cannot create a Class Mammal with individual squirrel in the Creatures ontology itself. That is an entity conflict.
2. I open Animals and under Active Ontologies … I import Creatures. I make Creature.squirrel the same individual as Mammal.squirrel, and Creature.dog the same individual as Mammal.canine. When I run my Reasoner, I observe that Mammal.canine now also belongs to the Creature class and hasBodySurface furry, inferred. Conversely, the Data Property hasGlossaryDefinition for Mammal.canine carried over to Creature.canine, however Mammal.canine now has duplicate instances of that Data Property.

Questions:
1. Is the duplicate Data Property a bug, and if not, is there anything I can assert to prevent it?
2. I am expecting that 20% of the synonyms (dog = canine) will have to be established by by business analysts through some tool, but are there better techniques than asserting a sameAs relationship for the 80% of terms that exactly match? I realize that can be risky, though since this experiment is over-simplified. In reality, each ontology has multiple classes and terms and there may be some homonyms (for example, squirrel and dog can also be verbs.)
3. When I go to exit my experiment, I’m asked if I want to save the changes I made to the ontologies within my workspace. I say no for now. What is the best practice? Should I create a new IRI for the merged ontologies and save that off (for ex, as RDF/XML) separately? If I do that, though, will there be entity conflicts such as what I observed in #1 above?

Double Data Property
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Questions about Data/Object Property Inferences between 2 Ontologies

sizedam
Could anyone kindly explain if the duplicate inferred data-property is a bug? See pic.

sizedam wrote
I am preparing to reconcile 2 different vocabularies within the same domain, but from 2 different organizations. I put together a little experiment to help decide whether to merge the vocabularies into 1 ontology or keep them separate and compare them via import. It is mostly working as I expected, but I have some questions.

Creatures Ontology:
* Class BodySurface with individuals furry & scaly
* Object Property hasBodySurface
* Class Creature with individuals squirrel & dog, each hasBodySurface furry

Animals Ontology:
* Data Property hasGlossaryDefinition
* Class Animal with sub-class Mammal, with individuals squirrel & canine, each with appropriate GlossaryDefinition values

Observations:
1. I cannot create a Class Mammal with individual squirrel in the Creatures ontology itself. That is an entity conflict.
2. I open Animals and under Active Ontologies … I import Creatures. I make Creature.squirrel the same individual as Mammal.squirrel, and Creature.dog the same individual as Mammal.canine. When I run my Reasoner, I observe that Mammal.canine now also belongs to the Creature class and hasBodySurface furry, inferred. Conversely, the Data Property hasGlossaryDefinition for Mammal.canine carried over to Creature.dog, however Mammal.canine now has duplicate instances of that Data Property.

Questions:
1. Is the duplicate Data Property a bug, and if not, is there anything I can assert to prevent it?
2. I am expecting that 20% of the synonyms (dog = canine) will have to be established by by business analysts through some tool, but are there better techniques than asserting a sameAs relationship for the 80% of terms that exactly match? I realize that can be risky, though since this experiment is over-simplified. In reality, each ontology has multiple classes and terms and there may be some homonyms (for example, squirrel and dog can also be verbs.)
3. When I go to exit my experiment, I’m asked if I want to save the changes I made to the ontologies within my workspace. I say no for now. What is the best practice? Should I create a new IRI for the merged ontologies and save that off (for ex, as RDF/XML) separately? If I do that, though, will there be entity conflicts such as what I observed in #1 above?

Double Data Property
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Questions about Data/Object Property Inferences between 2 Ontologies

sizedam
including larger clear pic
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Questions about Data/Object Property Inferences between 2 Ontologies

Csongor Nyulas
Administrator
I am not sure if it is a bug per se. It just shows that you have
asserted explicitly in your ontology something that could be inferred
automatically from other existing statements in your ontology. Seeing
this provides you with an opportunity to fix/simplify your ontology,
i.e. you can remove either the data property assertion on the 'canine'
individual, or the more generic statement that entail the same thing (to
find the source that contributed to the inferred axiom, click on the
question mark icon on its right).

So, I would say it is a feature, rather than a bug.

Csongor

On 04/21/2017 03:58 AM, sizedam wrote:

> including larger clear pic
> <http://protege-project.136.n4.nabble.com/file/n4668245/DoubleDataPropertyInference.png>
>
>
>
> --
> View this message in context: http://protege-project.136.n4.nabble.com/Questions-about-Data-Object-Property-Inferences-between-2-Ontologies-tp4668144p4668245.html
> Sent from the Protege User mailing list archive at Nabble.com.
> _______________________________________________
> protege-user mailing list
> [hidden email]
> https://mailman.stanford.edu/mailman/listinfo/protege-user


_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Questions about Data/Object Property Inferences between 2 Ontologies

Lorenz B.
@Csongor: Is this also the expected behavior for other inferred statements that are already asserted?

I am not sure if it is a bug per se. It just shows that you have asserted explicitly in your ontology something that could be inferred automatically from other existing statements in your ontology. Seeing this provides you with an opportunity to fix/simplify your ontology, i.e. you can remove either the data property assertion on the 'canine' individual, or the more generic statement that entail the same thing (to find the source that contributed to the inferred axiom, click on the question mark icon on its right).

So, I would say it is a feature, rather than a bug.

Csongor

On 04/21/2017 03:58 AM, sizedam wrote:
including larger clear pic
<http://protege-project.136.n4.nabble.com/file/n4668245/DoubleDataPropertyInference.png>



--
View this message in context: http://protege-project.136.n4.nabble.com/Questions-about-Data-Object-Property-Inferences-between-2-Ontologies-tp4668144p4668245.html
Sent from the Protege User mailing list archive at Nabble.com.
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user


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

-- 
Lorenz Bühmann
AKSW group, University of Leipzig
Group: http://aksw.org - semantic web research center

_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: Questions about Data/Object Property Inferences between 2 Ontologies

sizedam
@Consogner @ Lorenz - Thanks for taking a look at this!

I’ve isolated the problem in a much simpler test in a subsequent post “Duplicate Inferred Functional Data Property - Reasoner Bug?”.

Basically, with just 1 class, 1 individual, and 1 Data Property with as many restrictions I could think to apply, and no other relationships upon which to draw inferences, the Hermit/Pellet reasoners STILL inferred a duplicate Data Property for canine … so at this point I’m leaning towards this being more of a bug than a feature; it seems like a missing check in the logic which will lead to teleological, useless inferences. Note that I have not observed this behavior with Object Properties, only Data Properties.

Lorenz B. wrote
@Csongor: Is this also the expected behavior for other inferred statements that are already asserted?
Loading...