W3C home > Mailing lists > Public > semantic-web@w3.org > September 2010

Re: SPARQL Endpoint with Inference support

From: Andrea Splendiani <andrea.splendiani@bbsrc.ac.uk>
Date: Mon, 27 Sep 2010 14:19:40 +0000
Message-ID: <339406142106314@jngomktg.net>
To: Riccardo Tasso <tasso@elet.polimi.it>
Cc: "semantic-web@w3.org" <semantic-web@w3.org>
Yes, you can do that.
For updates, it may be more tricky.

In brief:

In Jena/Joseki you build up a model via a spec file.

Imagine your data files (the triplestore, not the rdf) are at
/exampled/dir/data,

you can build both a plain model which holds /example/dir/data or an an
infmodel, which has the same data and your choice of reasoners.

You can plug any of the two into Joseki (mutiple instances on different
ports)

Now:

if you use TDB as a Jena backend, I'm not sure it is safe versus concurrent
accesses (the two joseki accessing the same files). Not for sure in writing,
perhaps ok in reading.

You can use SDB (which is a relational backed triplestore): less performant,
but should support concurrent R/W access.

You can find more info on the Jena mailing list. If you check the archieves,
I recall some message about TDB/SDB and concurrency.

How big is your data ?

SDB il less performant than TDB, for big data (well... relatively) this can
make a difference.

ciao,
Andrea

Il giorno 27/set/2010, alle ore 16.09, Riccardo Tasso ha scritto:

> Grazie Andrea,
>     but what if I need to access data from a SPARQL endpoint? Could I 
> use Joseki to query two different models (one inferred, and one 
> "simple") such that they are based on the same data? Please note that 
> I'd like to have the possibility to execute updates too.
> 
> Cheers,
>     Riccardo
> 
> On 27/09/2010 15:19, Andrea Splendiani wrote:
>> Hi,
>> 
>> you can also use Jena for this.
>> You have a choice of reasoners which you can attach to the RDF model, in
>> your case you can use either a minimal "transitive reasoner" and (bits
of) a
>> rule based reasoner. You can select which rules you want to apply, or
choose
>> among default configurations.
>> If you define your own rules, use backward rules.
>> 
>> at the SPARQL level, this is completely transparent. But you can change
>> reasoners dynamically programmatically if you need.
>> 
>> ciao,
>> Andrea
>> 
>> Il giorno 22/set/2010, alle ore 12.06, Riccardo Tasso ha scritto:
>> 
>> 
>>> I'm interested in executing SPARQL queries over an endpoint, sometimes
>>> with reasoning, sometimes without (dynamically). The reasoning task I
>>> need are RDFS plus owl:sameAs. It is also an interesting question to me,
>>> wheter a SPARQL query could specify if the results should contain
>>> inferred triples or not.
>>> 
>>> Do you know a triple store, possibly supporting named graphs, suitable
>>> to my needs?
>>> 
>>> Thanks,
>>>     Riccardo
>>> 
>>> 
>>> 
>> Andrea Splendiani
>> Senior Bioinformatics Scientist
>> Centre for Mathematical and Computational Biology
>> +44(0)1582 763133 ext 2004
>> andrea.splendiani@bbsrc.ac.uk
>> 
>> 
>> 
>> 
> 

Andrea Splendiani
Senior Bioinformatics Scientist
Centre for Mathematical and Computational Biology
+44(0)1582 763133 ext 2004
andrea.splendiani@bbsrc.ac.uk
Received on Monday, 27 September 2010 14:20:50 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:38 GMT