On 09/07/12 13:45, Milorad Tosic wrote:
Hi,
I am aware of some earlier discussions on the list about blank nodes as well as
discussions on other places (e.g. [1]). So, my intention here is not to launch
a general purpose theoretic discussion here. Instead, I would focus on the
following question:
In the paper [2], authors say: "For example, ARQ is a commonly (re)used SPARQL processor
which supports a non-standard <_:b1> style syntax for terms in queries, indicating that
the term can only be bound by a blank node labeled “b1” in the data." However, a simple
query like the following
SELECT DISTINCT ?pp ?oo
WHERE {
<_:b1> ?pp ?oo .
This must the internal identifier,and "b1" is not that.
You need to get the real internal identifier:
1/ At the API, it's the AnonId from Resource.getId();
2/ To get the protocol engine to expose internal identifiers, ensure the
server is running with these context settings
Call
ARQ.enableBlankNodeResultLabels(true) ;
The TSV format output outputs (encoded) bnodes internal identifiers anyway.
}
gives an empty result set even though query
SELECT DISTINCT ?s ?pp ?oo
WHERE {
?s ?pp ?oo .
FILTER (isBlank(?s))
}
gives result set where _:b1 appears several times.
Irrelevent :-)
That "_:b1" is, I guess, the result-file-scoped translation, Normally
internal identifiers aren't exposed.
So, the question is whether I am using wrong syntax or something has changed in
ARQ or else?
What's your setup? How are you making the query?
Thanks,
Milorad
[1] http://www.w3.org/wiki/BnodeSkolemization
[2] web.ing.puc.cl/~marenas/publications/iswc11.pdf
