Ok, looks like we are getting somewhere! On Thu, May 19, 2011 at 3:38 PM, Enrico Franconi <franconi@inf.unibz.it>wrote: > > On 19 May 2011, at 22:30, David McNeil wrote: > > > Admittedly this is a long thread and it is hard to follow, but it seems > to me that Richard is advocating omitting the NULL values from the Direct > Mapping triples and that Enrico is ok with omitting the NULL values from the > Direct Mapping, so maybe that is an approach that would have broad support? > (This assumes that the Direct Mapping includes the schema information which > it sounds like the Direct Mapping editors agree that it should). > > Indeed. Basically the schema information allows you to reconstruct the NULL > values - as if you had them directly (which is admittedly easier, I guess; > and this has been the choice of SQL). > Note that a consequence of this is that the translated data alone is not > meaningful anymore as an RDF graph in its solitude, but it should be coupled > with schema information *and* a recipe on how to cook all this stuff > together: you have to prescribe how queries should be translated in order to > give the same/correct answer. > +1 This means that we need to extend our Direct Mapping to translate the schema into RDFs/OWL. And if you want to get the Nulls, you will need to use Optionals in your sparql query. > cheers > --e. >
Received on Thursday, 19 May 2011 20:46:07 UTC