I just evaluated my OLWIM-Lite repository performance and the results made me wonder a little...
A simple query such as:
This query runs faster with reasoning level on "owl-horst", "owl-max", "owl-horst-optimized" than the query given below using "empty" ruleset for reasoning, thus using OWLIM-Lite as plain RDF-repository.
This query I used with "empty" ruleset:
I find it a little strange - or am I missing something?
I read about sameAs-optimization here, but it is mentioned in context of OWLIM-SE. So it this also practiced at OWLIM-Lite or why can a query without reasoning be slower than a query, which requires reasoning?
Maybe some of you clever guys knows an answer to this.
You are quite right, that there is no special handling of owl:sameAs in OWLIM-Lite. However, I don't think a comparison of the execution speeds of two different queries in two different set-ups makes much sense (I'm trying to avoid the apples and oranges analogy). What would be more interesting to know is whether the same query runs faster with inference compared to without inference.
Some indication of the actual speeds might make it clear why you raise the question, i.e. some numbers.
All the best!
answered 22 Nov '12, 07:25