[Pellet-users] Expectations about Performance?

[James Howison]

Hi,

I'm using Pellet to build an inference model by reasoning with an OWL- 
DL ontology and a sample RDF data-file which contains about 70  
individuals.  Currently I'm finding that the reasoning to expand the  
model is taking upwards of an hour.  Given that I'm hoping to scale to  
10,000s of individuals this is a problem :)

I have control over the ontology and the reasoner configuration, so  
I've had a look through the list archives and tried some of the things  
suggested (removing cardinality restrictions, trying  
REALIZE_INDIVIDUAL_AT_A_TIME = false, I'm not sure if I can use the  
UNA) but that hasn't had much impact.

I wonder if anyone could be so kind as to help me understand whether  
there are things I can do to speed this up dramatically, or whether  
this is the performance I should expect.  I'm just learning about OWL  
and RDF, so it is quite possible I may have made some foolish  
mistakes.  I'm by no means wedded to the Jena interface, but I doubt  
that is the issue.  My immediate need is to run SPARQL queries like "? 
url rdf:type fc:Event" against the expanded model.

Here is a zip with the java code I'm running (via Jena)  
(ReasonOnRDFModel), the OWL schema (flosscomms.owl) and sample RDF  
data (as RDF/XML) (gaim-cabal.rdf).  You'd just need to set the path  
PREFIX to the files (I've marked that FIXME in the file).

http://freelancepropaganda.com/temp/pellet-testcase.zip

Thanks,
James