[Pellet-users] Concurrent Use of Pellet - Some Questions

[Paolo Bussotti]

Hi everybody.

With reference to the snippet in the Pellet FAQ about the concurrent use 
of Pellet,
which I have included below:

Model model = ModelFactory.createOntologyModel( 
PelletReasonerFactory.THE_SPEC, model );
model.read( "http://example.org/ontology" );
 /* ... any additional model modification ... */
((PelletInfGraph) model.getGraph()).classify();
model.query( ... );

I would like to know why the graph.classify() is used instead of , for 
example, graph.prepare() .
In facts -correct me if I'm wrong -both  implies kb.prepare().
So the focus seems to me to be addictionally calling also for 
classification.  (i.e. to  kb.classify  through  graph.classify()).

I have two questions:
1) If I use an external synchronisation construct allowing threads to 
access the pellet-backed model  in a mutually exclusive fashion,
    both for writing and querying, isn't  this enough to  ensure a  
correct behaviour?
2) Any externally synchronized thread prepares the graph (or the model) 
after model manipulation, and therefore the implied call to kb.prepare() 
should be done.
    So I think that any successively synchronized thread will start with 
a kb in the proper state, with changes reset via the kb.prepare() 
invocation.
3) If I make changes from the pellet-backed Jena model, prepare it, call 
for consistency check and possibly query it both through the model calls 
and SPARQL queries, it appears to me from the sources  that 
classification/realization  is done at query time, at least for many 
programmatic query. Also SPARQL queries provide correct results and 
appear to me (from the required time) to have a similar behaviour. I've 
worked with Pellet 1.4. Is it still true for the more recent version? If 
it is why then, use graph().classify() each time?           

I ask you this because I have to frequently modify  the model and  
prepare it  at the graph level in order to check consistency. 
Therefore, I imposed an external synchronization. Moreover I classified 
and realise explicitly the model only at the very start, relying to 
classification
occurring automatically when querying the model.

I thank you in advance,  and  hope to not to have formulated my question 
in a comprehensible way :-) .
If not, please advise me and I'll try to be more clear.   

Best Regards
        Paolo Bussotti