Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-owl] Difference between Pellet reasoner instances

Marius mgwiazda at onet.eu
Wed Jan 9 14:14:38 PST 2008


Hi Tania,

Could You explain why have you changed the Jena reasoner implementation not
to return indirect types? Isn't better to enable in OWL-API reasoner
implementation returning indirect types?
I'm asking because this is important feature for me.

Thanks for help,
Mariusz 



Tania Tudorache wrote:
> 
> Hi Mariusz,
> 
> Thanks for the details.
> 
> mentat wrote:
>> Hi Tania,
>>
>>
>> The difference is only between indirect types. Jena reasoner
>> implementation returns more indirect types for individuals than OWL-API
>> implementation. Look at individual "Jarek_Xksinski" (direct asserted to
>> class "Maz").
>>   
> Yes, I've checked and the Jena reasoner returns the indirect types as 
> well, while the OWL-API implementation only returns the direct types. 
> I've changed the Jena reasoner implementation to return only the direct 
> types. The fix will be available in the beta from January, not the one 
> from today.
> 
>> For me it looks like this  SWRL rule is fired. But maybe I'm wrong.
>>
>>
>>   
> 
> You are right! It was a big amazement for me. The Jena reasoner does 
> send the SWRL rules to the reasoner and the Pellet seems to interpret 
> them correctly. This has to be checked with the Pellet people, but the 
> results are encouraging. At the time when I implemented the reasoner 
> connection through Jena, the SWRL used to import the swrl.owl and 
> swrlb.owl. At that time, the SWRL rules where filtered out before 
> converting the Protege-OWL to a Jena model based on the imports. Later, 
> we have changed the way SWRL worked, and the imports of swrl.owl and 
> swrlb.owl were not necessary anymore. The side effect of this was that 
> the SWRL rules were not filtered anymore when the Protege model was 
> converted to a Jena model. But, I did not notice this, because the Jena 
> reasoner is not enabled in the UI. I have done most of the tests with 
> the OWL-API reasoner. Thanks for pointing this out.
> 
> Tania
> 
> 
>> Thank for help,
>> Mariusz
>>
>>
>>
>>
>>
>> Tania Tudorache wrote:
>>
>>   
>>> Marius,
>>>
>>> Thank you for the ontology. Can you tell me for which individual you get 
>>> different results?
>>>
>>> Tania
>>>
>>>
>>>
>>> mentat wrote:
>>>   
>>>     
>>>> Hi Tania,
>>>>
>>>> Thank You for reply.
>>>>
>>>>
>>>> Tania Tudorache wrote:
>>>>
>>>>     
>>>>       
>>>>> Hi Marius,
>>>>>
>>>>> Marius wrote:
>>>>>  
>>>>>       
>>>>>         
>>>>>> I'm using Protege-OWL API ver. 3.4 build 122. I have a questions:
>>>>>>
>>>>>> 1) What is difference between Pellet reasoner implementations
>>>>>> (ProtegePelletJenaReasoner.class and 
>>>>>> ProtegePelletOWLAPIReasoner.class)?
>>>>>>
>>>>>>       
>>>>>>         
>>>>>>           
>>>>> They do the same thing, but in different ways. The Jena 
>>>>> implementation was the first one and the way it works is that it 
>>>>> converts a Protege-OWL model to a Jena model and it attaches the 
>>>>> Pellet reasoner to it. The OWL-API implementation converts the 
>>>>> Protege-OWL model to an OWL-API model and uses the Pellet connection 
>>>>> available in OWL-API.
>>>>>
>>>>>  
>>>>>       
>>>>>         
>>>>>> When I use ProtegePelletOWLAPIReasoner.class I get different results 
>>>>>> than
>>>>>> when I use ProtegePelletJenaReasoner.class (reasoning about 
>>>>>> individuals is
>>>>>> more complex when using ProtegePelletJenaReasoner.class especially 
>>>>>> about
>>>>>> inferred types).       
>>>>>>         
>>>>>>           
>>>>> So, you get different results when computing the inferred types with 
>>>>> the Jena reasoner and with the OWL-API reasoner? I can imagine that 
>>>>> the difference is only in the direct types versus indirect types, but 
>>>>> I may be wrong.
>>>>> I could look at this if you send me the ontology and tell me for 
>>>>> which instance you get the difference.
>>>>>
>>>>>   
>>>>>       
>>>>>         
>>>> Yes, the difference is only between indirect types. Jena reasoner 
>>>> implementation returns more indirect types for individuals than 
>>>> OWL-API implementation. If You have time, look at my sample ontology 
>>>> at "Jarek_Xksinski" instance (direct asserted to class "Maz").
>>>>
>>>>
>>>>     
>>>>       
>>>>>> 2) Can this reasoners run SWRL rules?       
>>>>>>         
>>>>>>           
>>>>> The SWRL rules are not converted in the current reasoner 
>>>>> implementations. We will have this feature soon for the OWL-API 
>>>>> reasoner.
>>>>>   
>>>>>       
>>>>>         
>>>> Are you sure that Jena reasoner implementation doesn't fire SWRL 
>>>> rules? In attached ontology is a class "Spadkodawca". Definition of 
>>>> this class is: "Osoba AND spadekOtwarto SOME xsd:dateTime". On this 
>>>> ontology is also SWRL rule: "umarlDnia(?x, ?y) → spadekOtwarto(?x,
>>>> ?y)".
>>>>
>>>> Instance "Jarek_Xksinski" has property "umarlDnia" but not property 
>>>> "spadekOtwarto". Jena implementations returns that inferred type of 
>>>> this instance is "Spadkodawca",
>>>> OWL-API doesn't return this information. For me it looks like this 
>>>> SWRL rule is fired. But maybe I'm wrong.
>>>>
>>>>
>>>>     
>>>>       
>>>>>> 3) In protege wiki
>>>>>> http://protegewiki.stanford.edu/index.php/ProtegeReasonerAPI I read:
>>>>>> "Each of the above reasoner implement the ProtegeReasoner interface. 
>>>>>> If you
>>>>>> would like to use a direct connection to Pellet in your application, 
>>>>>> it is
>>>>>> recommended that you use the ProtegePelletOWLAPIReasoner"; but why?
>>>>>>
>>>>>>       
>>>>>>         
>>>>>>           
>>>>> The Jena implementation was more a proof of a concept that it is 
>>>>> possible to use any Jena reasoner on the converted Jena model. It is 
>>>>> also more memory intensive than the OWL-API implementation.
>>>>>
>>>>> If you don't need any of the Protege 3 plugins, I recommend that you 
>>>>> use the OWL-API and Protege 4 directly.
>>>>>
>>>>> Tania
>>>>>
>>>>>
>>>>>   
>>>>>       
>>>>>         
>>>> Unfortunately I'm using plugins (especially JessTab). But thank you 
>>>> for advise.
>>>>
>>>>     
>>>>       
>>>>>> Thank for help,
>>>>>> Mariusz
>>>>>>       
>>>>>>         
>>>>>>           
>>>>> _______________________________________________
>>>>> protege-owl mailing list
>>>>> protege-owl at lists.stanford.edu
>>>>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>>>>
>>>>> Instructions for unsubscribing: 
>>>>> http://protege.stanford.edu/doc/faq.html#01a.03
>>>>>   
>>>>>       
>>>>>         
>>>> Thank for help,
>>>>
>>>> Mariusz
>>>>
>>>> ------------------------------------------------------------------------
>>>>
>>>> _______________________________________________
>>>> protege-owl mailing list
>>>> protege-owl at lists.stanford.edu
>>>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>>>
>>>> Instructions for unsubscribing:
>>>> http://protege.stanford.edu/doc/faq.html#01a.03 
>>>>   
>>>>     
>>>>       
>>> _______________________________________________
>>> protege-owl mailing list
>>> protege-owl at lists.stanford.edu
>>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>>
>>> Instructions for unsubscribing:
>>> http://protege.stanford.edu/doc/faq.html#01a.03 
>>>   
>>>     
>>
>> _______________________________________________
>> protege-owl mailing list
>> protege-owl at lists.stanford.edu
>> https://mailman.stanford.edu/mailman/listinfo/protege-owl
>>
>> Instructions for unsubscribing:
>> http://protege.stanford.edu/doc/faq.html#01a.03 
>>   
> 
> _______________________________________________
> protege-owl mailing list
> protege-owl at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/protege-owl
> 
> Instructions for unsubscribing:
> http://protege.stanford.edu/doc/faq.html#01a.03 
> 
> 

-- 
View this message in context: http://www.nabble.com/Difference-between-Pellet-reasoner-instances-tp14361015p14722869.html
Sent from the Protege - OWL mailing list archive at Nabble.com.




More information about the protege-owl mailing list