W3C home > Mailing lists > Public > public-rdb2rdf-wg@w3.org > April 2011

Re: Keeping R2RML free of Direct Mapping dependency (ISSUE-25)

From: Ivan Herman <ivan@w3.org>
Date: Tue, 26 Apr 2011 10:25:25 +0200
Cc: RDB2RDF WG <public-rdb2rdf-wg@w3.org>
Message-Id: <E6E261BE-EF36-4572-9CDE-F105B3792F27@w3.org>
To: David McNeil <dmcneil@revelytix.com>
David,

I would like to understand. The original ISSUE-25 lists these as the main arguments:

[[[
1. R2RML mappings are easier to understand (the Direct Mapping specification does not need to be referenced to understand a custom mapping)
2. R2RML is easier to implement without the ties to the Direct Mapping
3. it is easier to implement R2RML in a consistent manner (differences in interpretation of the Direct Mapping spec do not leak into the implementation of R2RML itself)
4. this keeps contentious issues related to automatic blank node generation and URI generation out of R2RML (otherwise these issues leak into R2RML via the Direct Mapping support)
5. this approach allows the W3C, users, and vendors the chance to gain more experience with R2RML mappings before attempting to define standard default mappings that are a part of R2RML
]]]

If the group decides not to make a reference to the Direct Mapping in in R2RML, then we still have to define what happens to those parts of the table that a specific R2RML file does not cover. In other words, we would have to define a default behaviour. I am not sure why that approach would be simpler than to say that, in such a case, the Direct Mapping applies. After all, the Direct Mapping represents the simplest possible mapping to RDF; wouldn't that lead to reinventing the wheel in R2RML? In other words, I do not really buy arguments #1 and #2. As for your #3: if such dicrepancies occurs, this is something we will have to check and amend. This is the WG's job. The same for #4: there has to be a decision that works for both sides.

Of course, you may propose a very radical solution for the default case: if something is not specified, it is dropped from the output. Ie, if for a given cell there is no specification in the R2RML mapping, then no RDF output is generated for that cell. That approach, of course, would simplify things and would cut the ties with the DM, but I do not find that approach very compelling...

Bottom line: I think we need more discussion on this, and I am not yet ready to close this ISSUE by accepting the proposal...

Ivan







On Apr 19, 2011, at 19:16 , David McNeil wrote:

> Today in the working group meeting we discussed "ISSUE-25: Including direct mapping constructs in R2RML mappings"[1]. The consensus on the call seemed to be that ISSUE-25 should be closed with the resolution that we will keep R2RML free from dependencies on the Direct Mapping. However, we wanted to give all of the group members a chance to comment on this before closing the item. So, please respond if you have objections, in particular it was noted that Souren had been advocating the approach of allowing the Direct Mapping constructs to be implicitly included in R2RML mappings.
> 
> [1] - http://www.w3.org/2001/sw/rdb2rdf/track/issues/25
> 
> Thank you.
> -David


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf
Received on Tuesday, 26 April 2011 08:24:25 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:00:23 UTC