On 02/08/11 18:10, Glenn Ammons wrote:
I would like to apply the RDFS reasoner over an entire TDB dataset; at
runtime, the dataset is often extended with new named graphs through
SPARQL/Update queries.
Is it possible to write an assembler specification that instructs,
say, Joseki to apply the reasoner to the entire dataset? The
assembler HOWTO
(http://jena.sourceforge.net/assembler/assembler-howto.html) explains
how to construct a _single_ model that includes a reasoner but I don't
see anything there on configuring a reasoner for _all_ models in a
dataset.
I suppose that the assembler specification could build up the dataset
from individual models, each of which applies the reasoner, but that
would be a _big_ specification that would have to be regenerated (and
Joseki restarted?) every time a named graph is added to the dataset.
Thanks.
--glenn
Glenn,
There isn't anything to apply an RDFS reasoner to each graph in a TDB
dataset dynamically.
You could pre-process any data before loading it so that the RDFS
inference is applied before loading (RDFS being simple enough that this
is possible).
If you are thinking of large scale, this has two advantages:
1/ The RDFS reasoner can use a lot of memory - a TDB dataset can be
bigger than RAM.
2/ Query and runtime can exploit the TDB layer directly, not run through
the inference layer.
A disadvantage is that it does not help with retraction.
See
http://openjena.org/wiki/RIOT#Inference
for a small, reduced RDFS that works on a triple stream.
Andy
