OK thnks, I see that Jena TDB/SDB won't fit for my needs. Is it possible
to use Jena with 4store, i.e. take advantage of 4store performance and
scalability and Jena's API ?
Martin
On 08/06/2013 01:18 PM, Andy Seaborne wrote:
On 05/08/13 23:01, Dave Reynolds wrote:
On 05/08/13 14:15, Martin Vassilev wrote:
Hi all,
I would like to ask you a few questions about Jena.
I read in the documentation that
*Note:* Although OWL version 1.1 is now a W3C recommendation, Jena's
support for OWL 1.1 features is limited
I assume you mean OWL 2, if I recall correctly that was called OWL 1.1
for a while before the full extent of the changes were apparent.
1. Can you tell me what exactly is not supported (or what is supported)
from the OWL 1.1 and what version of OWL is fully supported.
Jena provides convenience methods for the complete OWL 1 language. It
provides built in rule based inference for subsets of the OWL full
dialect of OWL 1 as described in [1]. Third party tools like Pellet
provide complete OWL DL implementations compatible with Jena.
Jena currently has no direct support for any of the OWL 2 extensions
beyond OWL 1. However, since OWL 2 can be fully encoded in RDF (the
normative syntax) and Jena can handle any RDF, then you can read and
write any OWL 2 ontology, you just need to do a little more work.
Certainly people have build commercial OWL 2 tools on top of Jena.
There is no built in inference support for any of OWL 2 beyond OWL 1. To
find out exactly what Pellet et all provide for OWL 2 you would need to
check with them directly.
2. Can Jena handle 2 or 3 billions of triples using TDB.
It depends.
There is no hard limit on TDB size around there so with a big enough
machine you should be able to load that. It is pushing the sizes where
things can get slow though. So whether query performance will be
adequate for you will depend on your data, your queries and your
machine. The only way to be sure is to try it.
It will take a long time to load and only be able to answer simple
queries (basically, look up a resource by URI or inverse functional
property and get the values of properties of that object).
3. Is it possible to switch TDB with SDB relatively easy ?
Relatively but SDB is not advised for new projects unless you have a
strong need for it and I very much doubt it would cope with 2-3bT.
SDB will not work at 2-3 billion triples.
If you design for SPARQL as your interface, you use a cluster store
like 4Store. It's not full SPARQL 1.1, but it's close, is open source
(GPLv2) and runs on multiple machines for scaling.
Dave
[1] http://jena.apache.org/documentation/inference/#owl
