The direct mapping is a set of rules. Q: What happens if I have a table? A: Create an IRI for every tuple and an IRI for the table and create a Table Triple <IRI for each tuple, rdf:type, IRI for the table> Q: What happens to an attribute? A: For each tuple, create an IRI, then create an IRI for the attribute and then create a Literal Triple: <IRI for the tuple, IRI for the attribute, value of that attribute I've always said that the direct mapping should be represented in R2RML. But the series of these "questions", are basically IF-THEN rules. And I think every programmer knows IF-THEN. But if I'm not wrong, Alexandre did not agree of having a R2RML represenation of the direct mapping. Juan Sequeda +1-575-SEQ-UEDA www.juansequeda.com On Mon, Jul 11, 2011 at 5:16 PM, David McNeil <dmcneil@revelytix.com> wrote: > I don't have a full appreciation for the history of the Direct Mapping > spec, but my view is that it would make sense to define the Direct Mapping > in terms of R2RML. The essence would be to create a TriplesMap & SubjectMap > for every table and a PredicateMap & ObjectMap for every column. Perhaps I > am not the intended audience, but I don't find the formalisms in the current > draft very accessible. > > -David >
Received on Monday, 11 July 2011 15:39:28 UTC