- From: Alexandre Bertails <bertails@w3.org>
- Date: Sun, 21 Nov 2010 10:12:20 -0500
- To: Ivan Herman <ivan@w3.org>
- Cc: Eric Prud'hommeaux <eric@w3.org>, Juan Sequeda <juanfederico@gmail.com>, public-rdb2rdf-wg@w3.org
[ snip ] > > I expect we'll all have our strategies for sparql2sql, and that while it's not up to RDB2RDF to mandate one, it is our job to make sure that our rdb2rdf mapping language enable sparql2sql. > > Or at least enables in the 80% of the cases... And some sort of a documentation of that would be in order I'm not sure of what you mean: * an rdb2rdf solution should cover 80% of the rdb cases? * 80% of the rdb instances seen through rdb2rdf should enable sparql2sql? I believe that any solution (direct mapping, datalog rules, r2rml) SHOULD be complete in the domain: RDB. At least, the Direct Mapping Eric and I wrote already gives you that: the proof is done in the type-checking of our proof-of-concept in Scala. The complexity/simplicity of a solution will make sparql2sql more or less easy. For example, it's obvious that the Direct Mapping will be easier than r2rml as the behavior of the mapping is not changed by any rule. With r2rml, the dual sparql2sql would have to take the rules into account. I'm not sure yet of the requirements on the rules to enable the query rewriting on the logical view of the database. Alexandre.
Received on Sunday, 21 November 2010 15:12:28 UTC