This example has since been superceded by the one in the pipeline
- Load all prerequisite reference data first: vocabularies, components and codelists (see the metadata directory for the specifics).
- Establish conventions that map from the terminology used in the file to the reference data.
- Run the prepare step on the input regional-trade.csv to produce a intermediate regional-trade.slugged.csv and regional-trade.slugged.csv-metadata.json
- Run the final csv2rdf step on those last two files to produce the Observations, DataStructureDefinition and ComponentSpecs:
gem install linkeddata
rdf serialize --input-format tabular regional-trade.slugged.csv --output-format ttl
This produces the output in regional-trade.ttl.
The Split pipelines involves multiple csvw pipelines that each create a different resource of the cube - i.e. one pipeline per rdf:type. This would allow distinct outputs to handled separately (loaded into distinct graphs etc). It also obviates the need to include json-ld in the metadata; instead of creating this during the preparation phase (and having it passed through the translation unchanhed) we can use the translation phase to create this directly.