[Pellet-users] serialize a classified ontology

[James Howison]

I don't know the problem domain that well, but in mine I have one time  
(very slow) reasoning (~3 million individuals) and many queries.  My  
solution was to store the realized (reasoned) ontology, then load that  
up for querying.  I did this via Jena with model.writeAll (using N- 
Triples for quick serialization), then loading back into a database  
backend, this time without a reasoner.

Of course this might not be an option for you; you may be wanting  
things over than just the 'fully reasoned' model; like an idea of what  
is inferred and what wasn't etc.

On Jun 25, 2008, at 1:45 PM, Antonio Sanchez Ruiz-Granados wrote:

> I am using Pellet as a reasoner for a planner, using a similar  
> approach to the one described in Sirin's thesis. I'd like to extend  
> that work using some case-based techniques. The problem is that  
> classifying the domain ontology usually takes more time that solving  
> the planning problem, i. e., I am solving very simple planning  
> problems in domains with several constraints. That's why I was  
> wondering if I can pre-process the domain ontology in some way.
>
> Anyway, thanks for answering so quickly :)
> Antonio.
>
>
> Michael Smith escribió:
>> On Wed, 2008-06-25 at 14:55 +0200, Antonio Sanchez Ruiz-Granados  
>> wrote:
>>
>>
>>> I've realized that Pellet needs a lot of time to classify some  
>>> complex ontologies, and I'd like to know if I can do that  
>>> operation only one time and not any time I load the ontology  
>>> again. I mean, I'd like to load the original ontology, classify  
>>> it, and serialize the classified ontology.
>>>
>>
>> Using the command line interface with -c RDF will serialize the  
>> inferred
>> class hierarchy, omitting all other axioms.
>>
>>
>>> Then if I need to work with that ontology again I would load not  
>>> the original one but the classified. The problem is that I need  
>>> Pellet to know that the ontology has been previously classified,  
>>> so it doesn't need to classify it again.
>>>
>>
>> If computing the classification hierarchy is the only reasoner  
>> service
>> you are interested in, using just the inferred hierarchy from my
>> previous comment will probably be sufficient - i.e., with no other
>> axioms, classification is always very fast.
>>
>> If you are using other reasoning services (e.g, consistency  
>> checking of
>> instance data) there isn't a clear solution.  At present you cannot
>> serialize Pellet's internal state post-reasoning, so the reasoning  
>> will
>> be repeated if the ontology is loaded into a new KnowledgeBase.
>>
>>
>
>
> -- 
> __________________________________________
>
> Antonio A. Sanchez Ruiz-Granados
> Dpto. de Ingeniería del Software e Inteligencia Artificial
> Facultad de Informática
> Universidad Complutense de Madrid
> e-mail: antonio.sanchez at fdi.ucm.es
> web: http://gaia.fdi.ucm.es/people/antonio/
>
> ___________________________________________
>
>
> _______________________________________________
> Pellet-users mailing list
> Pellet-users at lists.owldl.com
> http://lists.owldl.com/mailman/listinfo/pellet-users
> _______________________________________________
>
> Sponsored by Clark & Parsia, LLC http://clarkparsia.com/