Newbie question: unsatisfiablity & Protege

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

Newbie question: unsatisfiablity & Protege

mdw
I have a simple toy ontology that the Hermit reasoner is reporting as having
unsatisfiable classes.

Looking at an earlier  posting
<http://protege-project.136.n4.nabble.com/Inconsistent-Ontology-Questions-td4657808.html#a4657810>
, I see:

/If you click on a class which is unsatisfiable(marked red) you can see the
descriptions of it, usually in the panel bottom right. There is a
equivalence relation to owl:Nothing listed. Click on the question mark(?) to
get explanations for the entailment./

In the attached ontology, I have classes that appear in red under
"inferred", but I do not see an equivalance relation for these classes in
the description view and consequently no opportunity to click on the
question mark to get the explanations.  

baking2.owl
<http://protege-project.136.n4.nabble.com/file/t377031/baking2.owl>  On a
related note, are there better ways to determine causes for
unsatisfiability?





--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user
Reply | Threaded
Open this post in threaded view
|

Re: Newbie question: unsatisfiablity & Protege

Michael DeBellis-2
That seems odd.  I've never seen behavior like this where the reasoner highlights things in red but when you click on the "?" the explanations come up empty. I wonder if this is a bug in the newest version. I still have the old version on another machine and just for the heck of it I'll try loading your ontology there and see if I get better explanations. 

BTW, why are you importing the BFO ontology for an ontology on baking? I know some people would strongly disagree with me but IMO that's just adding a bunch of classes that are confusing and don't provide value. For a baking ontology I would stick with concepts that someone who bakes would understand not continuants and occurrents. 

Michael

On Fri, Mar 29, 2019 at 12:09 PM mdw <[hidden email]> wrote:
I have a simple toy ontology that the Hermit reasoner is reporting as having
unsatisfiable classes.

Looking at an earlier  posting
<http://protege-project.136.n4.nabble.com/Inconsistent-Ontology-Questions-td4657808.html#a4657810>
, I see:

/If you click on a class which is unsatisfiable(marked red) you can see the
descriptions of it, usually in the panel bottom right. There is a
equivalence relation to owl:Nothing listed. Click on the question mark(?) to
get explanations for the entailment./

In the attached ontology, I have classes that appear in red under
"inferred", but I do not see an equivalance relation for these classes in
the description view and consequently no opportunity to click on the
question mark to get the explanations. 

baking2.owl
<http://protege-project.136.n4.nabble.com/file/t377031/baking2.owl>  On a
related note, are there better ways to determine causes for
unsatisfiability?





--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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
|

Re: Newbie question: unsatisfiablity & Protege

Michael DeBellis-2
In reply to this post by mdw
I tried loading your ontology in the older version of Protege (5.2) and got the same results. The ontology is highlighted mostly in red (as usually happens when you have an inconsistency) but when I click on the explanations I get blank windows, the explanation screen pops up but there is nothing there.  The only thing I can suggest is to go back to previous versions of your ontology, if you have them, that didn't have a contradiction and re-build to where you are making sure to run the reasoner after every change. That way you will see which change caused the contradiction. I don't think you should have to do that though, I think this may be a bug in the explanation part of Protege. BTW, I'm using Pellet not Hermit so I don't think the issue has to do with the reasoner but rather the part of Protege that takes output from the reasoner and generates explanations. But I don't know the architecture of Protege (where the reasoner ends and the explanation module begins) so it could be there is just something really unusual in the ontology that both reasoners can't generate explanations for. 

Michael

On Fri, Mar 29, 2019 at 12:09 PM mdw <[hidden email]> wrote:
I have a simple toy ontology that the Hermit reasoner is reporting as having
unsatisfiable classes.

Looking at an earlier  posting
<http://protege-project.136.n4.nabble.com/Inconsistent-Ontology-Questions-td4657808.html#a4657810>
, I see:

/If you click on a class which is unsatisfiable(marked red) you can see the
descriptions of it, usually in the panel bottom right. There is a
equivalence relation to owl:Nothing listed. Click on the question mark(?) to
get explanations for the entailment./

In the attached ontology, I have classes that appear in red under
"inferred", but I do not see an equivalance relation for these classes in
the description view and consequently no opportunity to click on the
question mark to get the explanations. 

baking2.owl
<http://protege-project.136.n4.nabble.com/file/t377031/baking2.owl>  On a
related note, are there better ways to determine causes for
unsatisfiability?





--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
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
|

Re: Newbie question: unsatisfiablity & Protege

Balhoff, James Patrick
In reply to this post by mdw
The problem is that owl:topObjectProperty is asserted to have domain ‘recipe’ and range ‘cooking step’. topObjectProperty connects every individual to every other individual, so every thing in this universe is both a recipe and a cooking step. I would not use domain and range with topObjectProperty. This is kind of an unusual situation, so I guess the explanation algorithm doesn't account for it properly.

Best regards,
Jim


> On Mar 29, 2019, at 3:09 PM, mdw <[hidden email]> wrote:
>
> I have a simple toy ontology that the Hermit reasoner is reporting as having
> unsatisfiable classes.
>
> Looking at an earlier  posting
> <http://protege-project.136.n4.nabble.com/Inconsistent-Ontology-Questions-td4657808.html#a4657810>
> , I see:
>
> /If you click on a class which is unsatisfiable(marked red) you can see the
> descriptions of it, usually in the panel bottom right. There is a
> equivalence relation to owl:Nothing listed. Click on the question mark(?) to
> get explanations for the entailment./
>
> In the attached ontology, I have classes that appear in red under
> "inferred", but I do not see an equivalance relation for these classes in
> the description view and consequently no opportunity to click on the
> question mark to get the explanations.  
>
> baking2.owl
> <http://protege-project.136.n4.nabble.com/file/t377031/baking2.owl>  On a
> related note, are there better ways to determine causes for
> unsatisfiability?
>
>
>
>
>
> --
> Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
> _______________________________________________
> 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
|

Re: Newbie question: unsatisfiablity & Protege

Michael DeBellis-2
Well done! I looked at the properties but it never even occurred to me to look at  owl:topObjectProperty. I agree it's not a good idea to define a domain and range for that property. 

Michael

On Sun, Mar 31, 2019 at 8:29 AM Balhoff, Jim <[hidden email]> wrote:
The problem is that owl:topObjectProperty is asserted to have domain ‘recipe’ and range ‘cooking step’. topObjectProperty connects every individual to every other individual, so every thing in this universe is both a recipe and a cooking step. I would not use domain and range with topObjectProperty. This is kind of an unusual situation, so I guess the explanation algorithm doesn't account for it properly.

Best regards,
Jim


> On Mar 29, 2019, at 3:09 PM, mdw <[hidden email]> wrote:
>
> I have a simple toy ontology that the Hermit reasoner is reporting as having
> unsatisfiable classes.
>
> Looking at an earlier  posting
> <http://protege-project.136.n4.nabble.com/Inconsistent-Ontology-Questions-td4657808.html#a4657810>
> , I see:
>
> /If you click on a class which is unsatisfiable(marked red) you can see the
> descriptions of it, usually in the panel bottom right. There is a
> equivalence relation to owl:Nothing listed. Click on the question mark(?) to
> get explanations for the entailment./
>
> In the attached ontology, I have classes that appear in red under
> "inferred", but I do not see an equivalance relation for these classes in
> the description view and consequently no opportunity to click on the
> question mark to get the explanations. 
>
> baking2.owl
> <http://protege-project.136.n4.nabble.com/file/t377031/baking2.owl>  On a
> related note, are there better ways to determine causes for
> unsatisfiability?
>
>
>
>
>
> --
> Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
> _______________________________________________
> 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

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

Re: Newbie question: unsatisfiablity & Protege

mdw
Jim, Michael,

Thank you for your insights!  I did not intend to assign a domain & range to
owl:topObjectProperty; I must have done that inadvertently.  Thank you for
pointing out my mistake.



--
Sent from: http://protege-project.136.n4.nabble.com/Protege-User-f4659818.html
_______________________________________________
protege-user mailing list
[hidden email]
https://mailman.stanford.edu/mailman/listinfo/protege-user