Dear all,
I am trying to build a knowledge graph from a non-normalised CSV file. Therefore it requires the selection of distinct values to avoid duplicated triples.
Against a triple store I perform that kind of CONSTRUCT with a sub-query applying the DISTINCT selection. Tarql parses that formulation without error, but returns an empty knowledge graph (log below).
Is there something wrong with the CONSTRUCT? Or is there a different strategy to address non-normalised data with tarql?
Thank you for reading.
$ cat select.sparqlSELECT DISTINCT ?id
WHERE {}
$ tarql select.sparql data.csv-------
| id |
=======
| "1" |
| "2" |
| "3" |
-------
$ cat construct.sparql
CONSTRUCT {
?id rdf:type vcard:Individual .
}
WHERE {
SELECT DISTINCT ?id
WHERE {}
}
$ tarql construct.sparql data.csv
$