Re: Semantics of ZeroOrOnPath and ZeroOrMorePath of property paths

Hi Adrian,

The way you describe ZeroOrOnePath and ZeroOrMorePath aligns with my understanding of the SPARQL specification. If the paper you cite suggests different results, that sounds like a problem with their formalization.

Thanks,
Greg

On Jan 22, 2018, at 7:21 AM, Adrian <skubella@uni-koblenz.de> wrote:
> 
> Hi everyone,
> 
> I have a question about the semantics of ZeroOrOnePath and ZeroOrMorePath of property paths. What are the correct query results in the following two situations?
> 
> 
> 1) ZeroOrOnePath
> 
> Let's consider the following graph G consisting of one triple:
> <foaf:alice foaf:knows foaf:bob>
> 
> Now we want to answer the query Q1:
> SELECT * WHERE {<foaf:chris> <foaf:knows>? ?o}
> 
> Based on the definition of ZeroOrOnePath with one variable in the W3C recommendation [1]
> eval(Path(X:term, ZeroOrOnePath(P), Y:var)) = { (Y, yn) | yn = X or {(Y, yn)} in eval(Path(X,P,Y)) }
> Q1 would return {(?o, <foaf:chris>)} since it is not checked whether <foaf:chris> is contained in G.
> 
> In contrast to this papers like [2] ZeroOrOnePath are defined in such a way that Q1 would return an empty result set.
> 
> 
> 2) ZeroOrMorePath
> 
> Let's assume the following query Q2 over the graph G shown above:
> SELECT * WHERE {<foaf:chris> <foaf:knows>* ?o}
> 
> Based on the definition of ZeroOrMorePath with one variable in the W3C recommendation [1], the query Q2 would return {(?o, <foaf:chris>)} even though <foaf:chris> is not element of the graph. (The definitions and a more detailed explanation is at the end of this mail.) According to [2] this query would return an empty result set.
> 
> 
> Can someone please explain to me, whether a subject not occurring in the graph should be contained in the result sets of ZeroOrOnePath and ZeroOrMorePath or not?
> 
> Thank you very much for your help.
> 
> Best regards,
> Adrian
> 
> [1]https://www.w3.org/TR/sparql11-query/#PropertyPathPatterns <https://www.w3.org/TR/sparql11-query/#PropertyPathPatterns>
> [2]https://users.dcc.uchile.cl/~mromero/papers/ppath.pdf <https://users.dcc.uchile.cl/%7Emromero/papers/ppath.pdf>
> 
> 
> In [1]  ZeroOrMorePath with one variable is defined as
> eval(Path(X:term, ZeroOrMorePath(path), vy:var)) ={ { (vy, n) } | n in ALP(X, path) }
> and ALP is defined as:
> 
> ALP(x:term, path) =
>     Let V = empty multiset
>     ALP(x:term, path, V)
>     return is V
> 
> # V is the set of nodes visited
> 
> ALP(x:term, path, V:set of RDF terms) =
>     if ( x in V ) return
>     add x to V
>     X = eval(x,path)
>     For n:term in X
>         ALP(n, path, V)
>         End
> 
> If I understand it correctly in the first iteration of ALP(x:term, path, V:set of RDF terms) x is added to V regardless of whether x is element of the graph or not. Thus, query Q2 would return {(?o, <foaf:chris>)} even though <foaf:chris> is not element of the graph.