For Apache Jena input: Conversion from CSV to RDF Format

Question

I am going to use Apache Jena and it takes RDF as the input format. But I've data in the CSV format. I researched a lot and couldn't find a way to convert it. Does anyone know how to do that efficiently.

I have gone thru tools like xml123 but the download link wasn't working.

Answer 1

Using jena-arq and jena-csv (both v3.0.1) the following way is working for me:

public static void main(String ... strings) throws Exception {
    CSV2RDF.init();
    //load through manager:
    //Model m = RDFDataMgr.loadModel("test.csv") ;
    //classic way to load:
    Model m = ModelFactory.createDefaultModel();
    try (InputStream in = JenaCSVTest.class.getResourceAsStream("/test.csv")) {
        m.read(in, "http://example.com", "csv");
    }
    m.setNsPrefix("test", "http://example.com#");
    m.write(System.out, "ttl");
}

The input (test.csv):

Town,Population
Southton,123000
Northville,654000

The output (rdf in turtle):

@prefix test:  <http://example.com#> .

[ test:Population  "123000"^^<http://www.w3.org/2001/XMLSchema#double> ;
  test:Town        "Southton" ;
  <http://w3c/future-csv-vocab/row>
          1
] .

[ test:Population  "654000"^^<http://www.w3.org/2001/XMLSchema#double> ;
  test:Town        "Northville" ;
  <http://w3c/future-csv-vocab/row>
          2
] .

See official doc jena-csv

UPDATE:

Starting jena-3.10.0 jena-csv has been retired. The last jena-csv release is 3.9.0. Instead you can use any other csv2rdf converters. For example, tarql.

A quick demonstration example for com.github.tarql:tarql version v1.2 (obtained through jitpack.io - it seems, there is no maven-central release):

    Path file = Paths.get(JenaCSVTest.class.getResource("/test.csv").toURI());
    String base = "http://example.com#";
    Model m = ModelFactory.createDefaultModel().setNsPrefix("xsd", XSD.getURI()).setNsPrefix("test", base);
    Graph g = m.getGraph();
    CSVOptions op = new CSVOptions();
    op.setDefaultsForCSV();
    String query = "PREFIX test: <" + base + ">\n" +
            "PREFIX xsd: <" + XSD.getURI() + ">\n" +
            "CONSTRUCT {\n" +
            "  ?Row a test:Row;\n" +
            "    test:town ?town;\n" +
            "    test:population ?population;\n" +
            "} \n" +
            "WHERE {\n" +
            "  BIND (BNODE() AS ?Row)\n" +
            "  BIND (xsd:string(?Town) AS ?town)\n" +
            "  BIND (xsd:integer(?Population) AS ?population)\n" +
            "}";
    TarqlQuery q = new TarqlQuery(QueryFactory.create(query));
    InputStreamSource src = InputStreamSource.fromFilenameOrIRI(file.toUri().toString());
    TarqlQueryExecution qe = TarqlQueryExecutionFactory.create(q, src, op);
    qe.execTriples().forEachRemaining(g::add);
    m.write(System.out, "ttl");

This snippet will generate the following RDF:

@prefix test:  <http://example.com#> .
@prefix xsd:   <http://www.w3.org/2001/XMLSchema#> .

[ a                test:Row ;
  test:population  123000 ;
  test:town        "Southton"
] .

[ a                test:Row ;
  test:population  654000 ;
  test:town        "Northville"
] .

Answer 2

You can also use https://github.com/AtomGraph/CSV2RDF which builds generic CSV/RDF graph and then transforms it using SPARQL CONSTRUCT query. (Disclaimer: I'm the author)