How can you change the data type (or some values) in a set of triples?

rdf, sparql

Solution

With SPARQL 1.1 you can do this with a SPARQL Update command like so:

DELETE { ?s ?p ?o }
INSERT { ?s ?p ?o2 }
WHERE
{
  ?s ?p ?o .
  FILTER(datatype(?o) = xsd:datetime)
  BIND(STRDT(STR(?o), xsd:dateTime) AS ?o2)
}

Update Commands apply over specific graphs so you may have to issue this mutliple times adding a `WITH <graph>` to the start of the command for each named graph you need to correct.

Problem

Say that someone you know accidentally inserted some triples into an RDF database with a datatype of xsd:datetime instead of the proper xsd:dateTime. What's the simplest way to correct this? I know I can do something like this to find the bad data: ``` select * where { ?s ?p ?o. filter (datatype(?o)=xsd:datetime) } ``` And I could take those results, fix them up in a text editor, delete the bad ones and insert the good ones... but I have to believe/hope there is an easier way.

Original source