Hi everyone,
we are experiencing some unexpected behaviour in evaluating queries
containing the owl:sameAs predicate. In particular, we noticed that
"owl:sameAs" triples are not (always) returned as part of the query result.
We are using OWLIM 5.3 Build 5777 and are aware that there was fix for
owl:sameAs triples in this version (This version resolved our problem
partially).
Let me try to summarize the behaviour.
We have an OWLIM repository containing multiple triples of the form (ex:s
owl:sameAs ex:other), in the particular scenario these are links from
DBpedia 3.8 files. We verified that the triple size of the repository as
given by the repository#size() and properties file are correct according to
the loaded data, i.e. it seems that the triples are physically available in
the repository.
Now, coming to the somewhat unexpected behaviour:
Assume I want to evaluate the boolean query "ASK { ?x <
http://www.w3.org/2002/07/owl#sameAs> ?o }" against this repository (which
is expected to return "true" according to our data). Then for different
settings I get the following results:
SPARQLRepository => false
HTTPRepository (includeInferred=false) => true
HTTPRepository (includeInferred=true) => false
Workbench (includeInferred=false) => true
Workbench (includeInferred=true) => false
Using the special named graph in the query, I always get "true", i.e. "ASK
FROM <http://www.ontotext.com/explicit> { ?x <
http://www.w3.org/2002/07/owl#sameAs> ?o }"
Can someone please comment on this issue? For me in particular the
SPARQLRepository case is relevant, as I do not have the chance to set
"includeInferred" to false there at all. Is there a possibility in OWLIM to
set the default value used?
Thanks a lot and best regards,
Andreas
_______________________________________________
Owlim-discussion mailing list
Owlim-discussion@ontotext.com
http://ontomail.semdata.org/cgi-bin/mailman/listinfo/owlim-discussion
