- From: Sören Auer <auer@informatik.uni-leipzig.de>
- Date: Mon, 14 Mar 2011 17:51:53 +0100
- To: RDB2RDF WG <public-rdb2rdf-wg@w3.org>
Hi all, As per our discussion during the last telco I wanted to draw our attention again on the Linked Data aspects of R2RML [1]. We should definitely not repeat the Linked Data principles in R2RML, but make clear, how certain Linked Data aspects can be dealt with and where problems would potentially occur. I see in particular the following three issues: 1. *identifier generation* - three possible approaches were outlined in [2], from my POV in particular the first ("Direct representation of entity references in database columns") and and third one ("Defining an entity mapping table") make sense, since they are easy to implement and would also allow reverse translations. 2. *namespaces* - I could not find any possibility to declare namespaces and respective prefixes in the current draft. is that left out intentionally? If not, we should add a respective declaration and make clear, that only namespaces which are governed by the user creating the mapping can be made de-referencable later on for Linked Data publishing. 3. *blank nodes* - we should make clear, that wherever blank nodes might be generated by any R2RML feature, this will result that the respective resources can not be directly de-referenced. There might be more aspects to be considered in order to make R2RML Linked Data proof, but this is what comes to my mind for now. Maybe others can comment on that. After we have a clear picture, we can think how this can be best reflected by the R2RML draft. Best, Sören [1] http://www.w3.org/2001/sw/rdb2rdf/wiki/LinkedDataAspects [2] http://www.w3.org/2001/sw/rdb2rdf/wiki/Entity_disambiguation
Received on Monday, 14 March 2011 16:52:12 UTC