- From: Alexander De Leon <me@alexdeleon.name>
- Date: Wed, 2 Jun 2010 13:22:58 +0200
- To: RDB2RDF WG <public-rdb2rdf-wg@w3.org>
Hi all, I just want to give my two cents contribution to the direct mapping / default mapping discussion. I see the advantages of being able to automatically generate a default R2RML mapping from the RDB schema. However, this is procedural mechanism which cannot be part of the specification of the R2RML (declarative) language. We as a WG can publish a separate recommendation / standard of how to do this. However, our main goal is to produce the specification of a mapping language which is expressive enough to cover most Relational-to-RDF mapping situations. I think that we should concentrate now in the specification of R2RML and not worry about the generation of the default mapping until the R2RML specification is fairly complete. Furthermore, as Juan's survey demonstrate, there are different approaches for the generation of the default mapping. I think that having this variety is healthy. I agree with providing some minimal guidelines on how to produce the default mapping, but we should leave it open for implementors to extend it with their own approaches. For instance, one can cover database vendor specific features in the automatic mapping generation, or one can also implement a semi-automatic approach for generating the default mapping. Cheers, Alex
Received on Wednesday, 2 June 2010 11:23:34 UTC