- From: Adrian <skubella@uni-koblenz.de>
- Date: Thu, 14 Dec 2017 16:06:07 +0100
- To: public-sparql-dev@w3.org
- Message-ID: <a7b5f49e-e3ca-9051-e4ed-0a26aff6abaa@uni-koblenz.de>
Hello,
I have a question about the semantics of ZeroOrOnePath of property paths.
In [1] ZeroOrOnePath with two variables is defined as:
eval(Path(X:var, ZeroOrOnePath(P), Y:var)) =
{ (X, xn) (Y, yn) | either (yn in nodes(G) and xn = yn) or {(X,xn), (Y,yn)} in eval(Path(X,P,Y)) }
Let's consider the following example graph G consisting of the following two triples:
<foaf:alice foaf:knows foaf:bob>
<foaf:bob foaf:knows foaf:bob>
Now we want to answer the following query:
SELECT * WHERE {?s <foaf:knows>? ?o}
Based on the former definition, this query would produce the following results:
1) {(?s, foaf:alice), (?o, foaf:alice)}, since foaf:alice is in nodes(G)
2) {(?s, foaf:alice), (?o, foaf:bob)}, since <foaf:alice foaf:knows foaf:bob> is in G
But {(?s, foaf:bob), (?o, foaf:bob)} is*no* result, since foaf:bob is in nodes(G)*and* <foaf:bob foaf:knows foaf:bob> is in G. This means the XOR condition in the definition of ZeroOrOnePath will be evaluate to false.
Based on my intuition of the semantics of ZeroOrOnePath I would have expected that {(?s, foaf:bob), (?o, foaf:bob)} is part of the result. Also in papers like [2] ZeroOrOnePath are defined in a way so that {(?s, foaf:bob), (?o, foaf:bob)} would be part of the result.
I would assume that there is a reason why the definition in [1] excludes {(?s, foaf:bob), (?o, foaf:bob)} from the result set. Could someone please explain me the reason, why {(?s, foaf:bob), (?o, foaf:bob)} was excluded?
Thank you very much for your help.
Best regards,
Adrian
[1]https://www.w3.org/TR/sparql11-query/#PropertyPathPatterns
[2]https://users.dcc.uchile.cl/~mromero/papers/ppath.pdf
Received on Thursday, 14 December 2017 15:59:39 UTC