Hi Holger,
In TBC I did not implement option 1 due to performance reasons. If we
would have to distinguish same objects, then we would need to walk
through a complete result set, while if we just accumulate statistics
then it's enough to just ask the triple store for the number of
Paul,
the count algorithm does not use SPARQL, but instead relies on the
Jena statistics providers on the Graph level. Walking through all
results and filtering them would be too slow. Also, a namespace is not
a safe criteria because any file may contain definitions from the same
Hi Bernard,
one might argue both directions. On the one hand side, same instances
should be suppressed (if you interpret the count as the number of
distinct instances. On the other hand, the count may be the number of
rdf:type triples for that class, in which case the current behavior is
On Apr 17, 2009, at 8:36 AM, Holger Knublauch wrote:
Hi Bernard,
Oops, sorry - I mean Hi Paul :)
Holger
On Apr 17, 2009, at 5:48 AM, PaulZ wrote:
I have been preparing a short showcase with Ensemble for a non
--~--~-~--~~~---~--~~
You received this