Speed of HermiT 1.3.3

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

Speed of HermiT 1.3.3

Tran Binh
hi all

I create an OWL file with Protege 4.1, it has:
134 Classes
69 Object properties
21 Data properties
1933 Individuals

then run HermiT 1.3.3 for reasoning, It takes more than 2 hours to finish :(

Is it normal for HermiT? is there any resolution to speed it up? It's seem useless if run too slow like that.

Cheers,
Binh
Reply | Threaded
Open this post in threaded view
|

Re: Speed of HermiT 1.3.3

Timothy Redmond
On 03/11/2011 07:24 AM, Tran Binh wrote:
> hi all
>
> I create an OWL file with Protege 4.1, it has:
> 134 Classes
> 69 Object properties
> 21 Data properties
> 1933 Individuals
>
> then run HermiT 1.3.3 for reasoning, It takes more than 2 hours to finish :(

There are three options you can look into.  First you have a choice of
reasoners.  Pellet and FaCT++ are the obvious ones.  If they all have
trouble that might mean that you have something in your ontology that is
difficult to analyze.  Though I use HermiT heavily, I switch between
reasoners all the time.

Second, the different reasoners have support e-mail addresses.  I know
that you can google for support e-mails for pellet and HermiT (on the
web page hermit-reasoner.com I think).  Try to include the ontology in
question - the size of the ontology is not enough to know where the
issues are.  Yours is pretty small though you do have more individuals
than most people have.

In addition, you do have some control over what the reasoner does when
it initializes (Click Reasoner->Configure) though this might not be easy
to figure out.  In the configuration panel, you can change the list of
inferences that Protege will display as it runs.  If you reduce the set
then the steps that the reasoner needs to take during initialization
will be smaller.

> Is it normal for HermiT? is there any resolution to speed it up? It's seem
> useless if run too slow like that.

The theory says that inference is intractable.  What is generally said
is that for most ontologies that people use in practice, reasoning can
be done in an efficient manner.  So there will be a dance between the
ontology developers and the reasoners.  As people use different OWL
constructs, the reasoners will run into different performance
bottlenecks and the reasoner developers will have to adapt.

-Timothy


> Cheers,
> Binh
>
> --
> View this message in context: http://protege-ontology-editor-knowledge-acquisition-system.136.n4.nabble.com/Speed-of-HermiT-1-3-3-tp3348353p3348353.html
> Sent from the Protege OWL mailing list archive at Nabble.com.
> _______________________________________________
> 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
Reply | Threaded
Open this post in threaded view
|

Re: Speed of HermiT 1.3.3

Yu Lin-2
In reply to this post by Tran Binh
Same problem with you.
My individuals are 290.
It takes around 1 hour.
I don't have patient for that.

Wait for answer....

Thanks,
Asiyah

--- Tran Binh <[hidden email]> wrote:

> hi all
>
> I create an OWL file with Protege 4.1, it has:
> 134 Classes
> 69 Object properties
> 21 Data properties
> 1933 Individuals
>
> then run HermiT 1.3.3 for reasoning, It takes more
> than 2 hours to finish :(
>
> Is it normal for HermiT? is there any resolution to
> speed it up? It's seem
> useless if run too slow like that.
>
> Cheers,
> Binh
>
> --
> View this message in context:
>
http://protege-ontology-editor-knowledge-acquisition-system.136.n4.nabble.com/Speed-of-HermiT-1-3-3-tp3348353p3348353.html
> Sent from the Protege OWL mailing list archive at
> Nabble.com.
> _______________________________________________
> 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
Reply | Threaded
Open this post in threaded view
|

Re: Speed of HermiT 1.3.3

Alan Ruttenberg-2
There are reasoner configuration controls. If you don't need all types of inference, e.g. Only consistency and classification, you might try adjusting those and seeing if they speed things up. Alternatively, try Fact++ or Pellet, which have different performance profiles.

-Alan

On Mar 12, 2011, at 6:38 AM, Yu Lin <[hidden email]> wrote:

> Same problem with you.
> My individuals are 290.
> It takes around 1 hour.
> I don't have patient for that.
>
> Wait for answer....
>
> Thanks,
> Asiyah
>
> --- Tran Binh <[hidden email]> wrote:
>
>> hi all
>>
>> I create an OWL file with Protege 4.1, it has:
>> 134 Classes
>> 69 Object properties
>> 21 Data properties
>> 1933 Individuals
>>
>> then run HermiT 1.3.3 for reasoning, It takes more
>> than 2 hours to finish :(
>>
>> Is it normal for HermiT? is there any resolution to
>> speed it up? It's seem
>> useless if run too slow like that.
>>
>> Cheers,
>> Binh
>>
>> --
>> View this message in context:
>>
> http://protege-ontology-editor-knowledge-acquisition-system.136.n4.nabble.com/Speed-of-HermiT-1-3-3-tp3348353p3348353.html
>> Sent from the Protege OWL mailing list archive at
>> Nabble.com.
>> _______________________________________________
>> 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
_______________________________________________
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