Dear all,

I am new to websemantic, and did only try some examples using the jena
api (and some Protege too). I also do read many useful posts here,
but still need a 'simple' overview of the current situation with
Jena2.

Here is a question to understand the general principle when it comes
to using inferenced models.

I understand that we start with an ontology (the schema), and some
data (the individuals). Than we load this in memory (from files or a
RDBMS). Than some engines are run to obtain a new inferenced model,
maybe based on some custom rules too. And than this inferenced model
can be queried.

I've read that all the inferenced models are handled in memory,
so the inferenced model is never saved back to the DB ?

If yes, is that the way that Jena does it or is it the way to do it in
semanticweb, no matter which tool you use (sesame, etc.) ?

Is it anyway better to query a model in memory then to query the DB ?

Well, than how do we handle huge models ? Will we have to decide which
part of the model to load and query ?

So, in practice, does it mean that a triple store has to be put and
kept in memory 'for ever'. And that only when a crash happens the all
process (loading, inference, etc.) will be done again ?

Thanks a lot for any help.
Fabian