Re: Not quite crossing the pons asinorum of construct

[Andy Seaborne]

On 29/12/10 13:15, Benson Margulies wrote:
because it's used in a UNION:

Let me see if I've got this:

?entity hasAnInstrument "Trumpet"
{ BIND(?entity AS ?s)}

My mistake - this will do nothing as ?entity isn't bound (SPARQL 
evaluation is bottom-up, functional evaluation, not as-you-read-it 
top-to-bottom of the query.

More deviousness needed to cover the 3 cases of no steps of owl:sameAs, 
forward owl;sameAs and backward owl:sameAs ....

CONSTRUCT{ ?s ?p ?o }
WHERE
{
    GRAPH x-arq:UnionGraph {

       { ?s rex:hasNormalizedText "Kerr" }
       UNION
       { ?entity rex:hasNormalizedText "Kerr" .  ?s owl:sameAs ?entity }
       UNION
       { ?entity rex:hasNormalizedText "Kerr" .  ?entity owl:sameAs ?s }
       ?s ?p ?o .
    }
}

but the property path way is still better, because it treats owl:sameAs 
as symmetric and transitive:

GRAPH x-arq:UnionGraph {
   ?entity rex:hasNormalizedText "Kerr" .
   ?entity (owl:sameAs|^owl:sameAs)* ?s
   ?s ?p ?o
 }

        Andy