Semantic Overflow

Question

2

I'm running SPARQL queries like:

prefix ch: <http://pele.farmbio.uu.se/chembl/?>

select count(?s) where {
  ?s a ch:Compound .
}

or any other 'a' query asking all Resources of a certain rdf:type, which is quite slow resulting in a time out... how can I get Virtuoso to be a bit speedier here? Can I index the store so that it more efficiently can return all resources rdf:type some:Foo?

flag	asked Jan 15 at 19:32 Egon Willighagen 620●2●10

Pēteris Caune · Answer 1 · 2010-01-18 11:11:01Z UTC

There are default indices on triples table (RDF_QUAD), they are only useful if you specify graph in SPARQL query. You can either:

add "FROM <...>" in your queries so default indices get used
create new indices that ignore graph. More info on this here: RDF Data Access and Data Management: Performance Tuning.

With proper indexing such simple query should execute quickly, or at very least it shouldn't timeout. If you need even more speed, you'll probably need to cache these aggregate counts somewhere.

hwilliams · Answer 2 · 2010-01-19 02:07:03Z UTC

3

Note you can also determine the indexes in place on your Virtuoso RDF_QUAD table using the STATISTICS call

answered Jan 19 at 2:07

hwilliams
41●1

How to speed up rdf:type foo:Bar SPARQL queries by Virtuoso tuning?

2 Answers