Re: Default Mapping vs Direct Mapping

On Mon, Oct 25, 2010 at 7:00 AM, Harry Halpin <hhalpin@w3.org> wrote:

> > Hi Everybody
> >
> > I've seen the use of Default Mapping and Direct Mapping in several
> places.
> > Soeren also pointed this out in his comments. So, what do these terms
> > mean?
> > Or are they the same?
> >
> > We need to come to a consensus asap!
>
> I think either term is OK as long as it is well-specified.
>
> >
> > IMO, a Direct Mapping is a mapping from a RDB to RDF where tuple pk are
> > subjects, attributes of a table are predicates and the values are objects
> > (or something like this... this will be clearly defined in the direct
> > mapping document). A RDB2RDF system should use the direct mapping as the
> > default mapping, meaning, when the user does not customize a mapping and
> > lets the RDB2RDF system expose the RDB data automatically as RDF, it will
> > do
> > it by the direct mapping.
>
> Agreed. The Direct Mapping is clearly Eric's algorithm.
>

I don't think it is anybody's algorithm. It simply converting a relation to
a graph. Tuple id is the subject, attribute is the predicate, value is the
object.

The result of the direct/default mapping is the consensus of this WG of how
to do that in RDF. Hence, we need to agree on as a group in order to
standarize: how to create uris, what to do when there aren't pks, etc....



>
> >
> > I'm guessing that R2RML document needs to reference the Direct Mapping
> > document. Can a FPWD reference a document that is not ready? I do not
> feel
> > comfortable about this.
>
> I'm OK in a FPWD in referring to an Editor's document.
>
>
> Also, one could thing would be to copy an English-language version of the
> algorithm from Eric's document and put it in R2RML. Therefore, if people
> *just* want to implement the direct mapping document, they can, but also
> we make sure that the direct mapping document and R2RML are on the same
> page.
>

This would be copying-pasting the first part of the direct/default document
to the R2RML then. Is this ok?

Per the meeting on thursday, Marcelo and I are working on a new version of
the doc based on what Eric has already done.


> >
> > Juan Sequeda
> > +1-575-SEQ-UEDA
> > www.juansequeda.com
> >
>
>