Re: R2RML mappings of Non-unique Tables

On Sat, May 12, 2012 at 11:33 PM, Eric Prud'hommeaux <eric@w3.org> wrote:

> I wrote up the issue and use cases for R2RML's handling of non-unique
> tables:
> <http://www.w3.org/2001/sw/rdb2rdf/wiki/Non-unique_Tables>
>

Eric - Thanks for writing this up. It is helpful.

I have a question for clarification. As a matter of fact, this statement
does not seem accurate to me:

"This makes it impossible for one to generate an R2RML mapping to the DM of
a table with potentially repeated rows."

An R2RML feature not mentioned in the writeup is the ability to define a
logical view in an R2RML mapping. Such a view can contain a SQL statement
(and R2RML has been defined with the idea of using database specific SQL in
such cases). I believe this capability means that we cannot say it is
"impossible" for an R2RML mapping to capture the DM of a table with
repeated rows. We can debate about whether it is practical to write or if
the runtime costs are too high, etc. but I don't think we can say it is
impossible.

-David