On 24/05/14 20:17, Qiaser Mehmood wrote:
Hey,
I got a strange situation while querying an endpoint by running jena
execConstruct (), OR execConstructTriples().
The issue is uncontrollable loop
Your query generates a large result - the whole database scaled up by
the number of classes.
You don't show your code but I guess you are using the default results
format, which is RDF/XML, which is slow to parse.
The total results from the remote are 91M. 2,226,435 lines of RDf/XML.
For only 742144 triples.
(WARN riot:77 - [line: 4, col: 41] {W108} Not an XML Name: ' ')
The result returned by the endpoint are not correct. (rdf:nodeID="" by
the look of it). Luckily, it's a WARNing not an ERROR. Unluckily,
it's sign of problem in the results
All the bnodes are the same bnode because they have the same rdf:nodeID.
by querying an endpoint http://roma.rkbexplorer.com/sparql/ and the actual
query i use to run is
PREFIX void: <http://rdfs.org/ns/void#>
> CONSTRUCT { <http://roma.rkbexplorer.com/sparql#dataset>
void:classPartition [ void:class ?c ; void:propertyPartition [
void:property ?p ] ] }
WHERE { ?s a ?c ; ?p ?o .}
which is the whole database multiplied by the number of classes.
Expensive and large.
It looks like the remote endpoint does not do full duplicate suppression
in graph results (it isn't require to) in the interests of streaming.
The graph returned is just 157 distinct triples.
the http query generated by jena QueryExecutionFactory.sparqlService(endpoint,
query) is as follow:
http://roma.rkbexplorer.com/sparql/?query=PREFIX++void%3A+%3Chttp%3A%2F%2Frdfs.org%2Fns%2Fvoid%23%3E%0A%0ACONSTRUCT+%0A++%7B+%3Chttp%3A%2F%2Froma.rkbexplorer.com%2Fsparql%2F%23dataset%3E+void%3AclassPartition+_%3Ac0+.%0A++++_%3Ac0+void%3Aclass+%3Fc+.%0A++++_%3Ac0+void%3ApropertyPartition+_%3Ac1+.%0A++++_%3Ac1+void%3Aproperty+%3Fp+.%7D%0AWHERE%0A++%7B+%3Fs+%3Chttp%3A%2F%2Fwww.w3.org%2F1999%2F02%2F22-rdf-syntax-ns%23type%3E+%3Fc+.%0A++++%3Fs+%3Fp+%3Fo%0A++%7D%0A
would you please solve this issue in jena library, since the loop starts within
jena library as execConstruct or execConstructTriples function is called. I am
unable to handle it in my java code since looping start internally.
Not a problem in Jena, which is just passing on your query to the remote
end and parsing the results.
Andy
Thanks.
Qaiser.
