[Pellet-users] SPARQL endpoint using pellet - for ActiveRDF Ruby on Rails use

James Howison james at howison.name
Sat Nov 29 03:44:06 UTC 2008


On 28 Nov 2008, at 6:12 PM, Silas Davis wrote:

> Hello,
>
> I will start with my specific desires:
>
> I want to use Pellet to reason over an OWL ontology in real-time,
> answering SPARQL queries requesting inferable information. I would  
> like
> to use ActiveRDF (http://www.activerdf.org/) for Ruby (and Rails), an
> 'Object  Description Logic Mapping' layer to query my ontology
> programmatically.
>
> ActiveRDF provides various 'adapters' that hook it up to datastores.  
> One
> such adapter is a SPARQL adapter that communicates with 'SPARQL'
> endpoints as described by:
>
> http://esw.w3.org/topic/SparqlEndpointDescription
>
> So far, I have been able to use ActiveRDF with the non-inferencing
> SPARQL endpoint provied by Joseki (http://www.joseki.org/).

You should be able to make an inferencing model back the Joseki  
store.  It is a matter of getting description in the Joseki config  
file right, using the Jena assembler language:

http://jena.sourceforge.net/assembler/

One ought to be able to have either one of the Jena reasoners or  
Pellet work that way (the Model will load into RAM and the reasoning  
will be performed on the first query, then cached.  The reasoning will  
need to be repeated each time Joseki is started up, and will need to  
"fit" into RAM.  If that's not feasible, and you aren't adding  
statements, then you could inference on a large machine, then save as  
a database backed model (SDB, TDB) and have Joseki expose that as a  
SPARQL endpoint.

You'll want to consult the jena-dev list, if you run into trouble with  
the assembler + Joseki combo.

> I would really like to run a server that acts as a proxy for Pellet,
> supporting the SPARQL endpoint standard. I know Pellet already  
> provides
> similar functionality (as a server) via DIG. One option would be to
> write a DIG adapter for ActiveRDF.
>
> I am not sure how widespread or accepted a standard SPARQL endpoints
> are, but perhaps this would be of more general use for people  
> wanting to
> exploit Pellet's reasoning abilities. Could anyone offer any  
> thoughts on
> this? Can I already accomplish my aims in another way (Jena?)? Any
> suggestions appreciated.

Always interested in other options too :)

--J


More information about the Pellet-users mailing list