Workflow for this idea:

1. Convert the data source to generic RDF
   • This must still be implemented manually per data type. For example, for CSV / relational data one can map rows to nodes, then connect its column values to it via predicates derived from column names.
2. Insert the generic RDF into a SPARQL query engine
3. Write CONSTRUCT queries to convert generic RDF into more semantic RDF
4. (Maybe add a loop around 2 and 3 to be able to perform multiple iterations of improvements to the RDF)