Reasoner errors

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

Reasoner errors

Aqualung
Hello all,

Attempting to classify ontology using both Pellet 1.5 and FaCT++ that come
with the latest Protégé 4.0 beta version (build 103). Get following error
messages respectively:

ReasonerException:
org.mindswap.pellet.exceptions.UnsupportedFeatureException: Axiom:
TransitiveObjectProperty(locatedIn)
 
org.protege.editor.owl.model.inference.OWLReasonerManagerImpl$1.run(OWLReaso
nerManagerImpl.java:247)
    java.lang.Thread.run(Unknown Source)

ReasonerException:
uk.ac.manchester.cs.factplusplus.owlapi.FaCTPlusPlusReasonerException:
uk.ac.manchester.cs.factplusplus.NonSimpleRoleInNumberRestrictionException:
http://www.ifomis.org/obo/ro/1.0#locatedIn
 
org.protege.editor.owl.model.inference.OWLReasonerManagerImpl$1.run(OWLReaso
nerManagerImpl.java:247)
    java.lang.Thread.run(Unknown Source)

Don't recall this happening w/build 102 or earlier (though I only used
Protégé in the past).

The property in question (locatedIn) is indeed transitive, and had been
declared as such for a long while. Removing the transitivity check yields a
consistent (thus Pellet classifiable) ontology.

Please help,

Christian



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

Re: Reasoner errors

Aqualung
I've been asked by a non-subscriber to post the following message,
presumably pertaining to my previous submission. Here goes:

> -----Original Message-----
> From: [hidden email] [mailto:[hidden email]]
> Sent: Monday, November 10, 2008 13:24
> To: [hidden email]
> Subject: INTERESTING!!!
>
> howdy, in addition to what you say i have attached two short ontologies
> (see attachment) that produce wicked results once they're checked by
> "pellet1.5" in protégé 4.0.  please see the results produced by
> "pellet1.5". please see what the authors of the following
> publication...
>
> http://www.webont.org/owled/2008dc/papers/owled2008dc_paper_20.pdf
>
> ...say about the reasoners ("page 3", last but one paragraph of
> "chapter 2").you'll be surprised to see it even in the body of
> literature now. apparently, there is no considerable maintenance of the
> reasoner source codes taking place (anymore). what do you think? ciao
> ciao, mat

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

Reasoner_madness_01.owl (6K) Download Attachment
Reasoner_madness_02.owl (6K) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: Reasoner errors

Bijan Parsia-3
In reply to this post by Aqualung
Christian,

On 10 Nov 2008, at 12:00, Cristian Cocos wrote:

> Hello all,
>
> Attempting to classify ontology using both Pellet 1.5 and FaCT++  
> that come
> with the latest Protégé 4.0 beta version (build 103). Get following  
> error
> messages respectively:
>
> ReasonerException:
> org.mindswap.pellet.exceptions.UnsupportedFeatureException: Axiom:
> TransitiveObjectProperty(locatedIn)

This is not a helpful error message.

> org.protege.editor.owl.model.inference.OWLReasonerManagerImpl$1.run
> (OWLReaso
> nerManagerImpl.java:247)
>     java.lang.Thread.run(Unknown Source)
>
> ReasonerException:
> uk.ac.manchester.cs.factplusplus.owlapi.FaCTPlusPlusReasonerException:
> uk.ac.manchester.cs.factplusplus.NonSimpleRoleInNumberRestrictionExcep
> tion:
> http://www.ifomis.org/obo/ro/1.0#locatedIn

This is a bit more helpful

> org.protege.editor.owl.model.inference.OWLReasonerManagerImpl$1.run
> (OWLReaso
> nerManagerImpl.java:247)
>     java.lang.Thread.run(Unknown Source)
>
> Don't recall this happening w/build 102 or earlier (though I only used
> Protégé in the past).
>
> The property in question (locatedIn) is indeed transitive, and had  
> been
> declared as such for a long while. Removing the transitivity check  
> yields a
> consistent (thus Pellet classifiable) ontology.

The problem is that transitive (i.e., "non simple") properties are  
not allowed to occur in number restrictions (in order to ensure  
decidability). In a prior version of protege 4, checking for that was  
disabled in Pellet (though not for FaCT++). Pellet has an "ignore  
problems  silently) mode which, obviously, isn't an appropriate  
default :)

Once the OWL 2 specs settle down, several people intend to build OWL  
syntax validators. I'm very confident such will be built into P4.

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: Reasoner errors

Bijan Parsia-3
In reply to this post by Aqualung
On 10 Nov 2008, at 13:31, Cristian Cocos wrote:

> I've been asked by a non-subscriber to post the following message,
> presumably pertaining to my previous submission. Here goes:
>
>> -----Original Message-----
>> From: [hidden email] [mailto:[hidden email]]
>> Sent: Monday, November 10, 2008 13:24
>> To: [hidden email]
>> Subject: INTERESTING!!!
>>
>> howdy, in addition to what you say i have attached two short  
>> ontologies
>> (see attachment) that produce wicked results once they're checked by
>> "pellet1.5" in protégé 4.0.  please see the results produced by
>> "pellet1.5". please see what the authors of the following
>> publication...
>>
>> http://www.webont.org/owled/2008dc/papers/owled2008dc_paper_20.pdf
>>
>> ...say about the reasoners ("page 3", last but one paragraph of
>> "chapter 2").you'll be surprised to see it even in the body of
>> literature now.

This person is being silly. That paper reports that the error is  
reported but also asks that better error messages/explanations be  
provided. There's a clear reason why to wait on this (since the  
standard keeps changing, though last call is soon), so a clear  
explanation why *this* bit of code has not been written/updated.

>> apparently, there is no considerable maintenance of the
>> reasoner source codes taking place (anymore). what do you think? ciao

And even sillier here. Pellet just went to 2.0rc. HermiT has been  
released. FaCT++ is continually updated. Etc. etc.

Basically, there's been a gap left since the old syntax validators  
stopped being useful (since they no longer cover the most common  
language) and the new ones being implemented (since the spec hasn't  
been stable). OWL 2 is *not yet standardized* and, indeed, has been  
in great flux.

Since all the reasoner and other tools implementors I know of are  
very responsive, I find such spreading of FUD to be somewhat  
obnoxious. In fact, only someone who doesn't pay any attention at all  
to the current churn would come to the "reasoners/tools aren't beign  
updated".

Furthermore, if one wishes more targeted support, several companies  
offer support contracts so that one can prioritize work.

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: Reasoner errors

Bijan Parsia-3
On 10 Nov 2008, at 14:15, [hidden email] wrote:

> thank you for your reply, bijan. where can i get pellet 2.0rc from?  
> give people a fair chance to be  obnoxious

I think you had more than your fair chance and you took it!.

> as nobody can be as splendid as you are.

Ooo, condescending sarcasm! But rather of low quality. I trust you'll  
understand that I'll be following your example only in spirit, rather  
than to the letter.

> in other words: don't be so offensive, bijan.

Let's see: You wrote something that was patently false (and easily  
verifiably so) and was a silly line of reasoning anyway (even if it  
were a bug, not all bugs, even trivial ones, get fixed on *your*  
schedule, for a variety of reasons).

(But really, if you had taken 10 seconds to google you would have  
found out that the reasoners in question are under active  
development. So to what purpose was your speculation.)

Me: I wrote that what you said was silly, which it was. I explained  
several ways that it was and gave a bunch of useful information. You,  
instead of apologizing, decided to take it to another level. Woohoo!

> which date exactly did the paper come out?

Ooo, yes, please, try to defend the indefensible. It's always a good  
move.

> is the info given there still valid? i'm not a computer expert or  
> description logic specialist. i'm new to this! so, how am i  
> supposed to know who is responsive and who isn't?

If you are that inexperienced, then how are you to know whether  
programs are being actively developed? Yet you had no problem  
speculating there.

In my experience, such claims are bullying moves (often careless ones).

> i wasn't born as an ontologist and so weren't you.

Neither of us was born gracious either, but somehow I've managed.

> so let me be silly. look at it from this point of view: otherwise  
> there would be no room for improvement.

Non sequitur.

> this means, be more considerate when talking to or about beginners

Most beginners, in my experience, don't start out by making  
indefensible and rather obnoxious claims. They ask questions.

> or you will find yourself regarded as being not very helpful

Feel free to regard me as not helpful.

> when comes to supporting the spread and proliferation of the  
> concept of ontology.

You're *so* right. I am a blight. It's unbearable how I hold back  
ontology technology! Mea culpa!

> thank you for your help anyway.

You're welcome.

> again, i'm just a beginner but i'm willing to learn,

Try learning *etiquette* first, at least before you go on a extended  
lecture on other people's manners.

> so be gentle. okay? thanks.

The sad thing was that I *was* gentle. You were being silly *and*  
obnoxious. I just commented on the silly and hoped you'd take the  
hint. Oh well.

Cheers,
Bijan.

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