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

Re: Request for comments: suggesting some minor R2RML changes

From: Eric Prud'hommeaux <eric@w3.org>
Date: Wed, 6 Jul 2011 08:20:42 -0400
To: Richard Cyganiak <richard@cyganiak.de>
Cc: W3C RDB2RDF <public-rdb2rdf-wg@w3.org>, Souripriya Das <souripriya.das@oracle.com>, David McNeil <dmcneil@revelytix.com>, Ivan Mikhailov <imikhailov@openlinksw.com>
Message-ID: <20110706122041.GA9011@w3.org>
* Richard Cyganiak <richard@cyganiak.de> [2011-07-06 09:45+0100]
> Below are a number of possible small changes/fixes to R2RML that we should consider. I think most of them are no-brainers, but would like to ask for feedback because they will require changes to the existing implementations. I'd appreciate some quick +1/-1 comments; if anything proves to be controversial, then I'll file it as an issue in the tracker for later discussion.
> Thanks,
> Richard
> ISSUE: capitalization of rr:SQLQuery property; according to conventions for predicates, it should be rr:sqlQuery

+1 (ericP)

> ISSUE: Why two classes rr:PredicateMap and rr:RefPredicateMap? They behave exactly the same, we could drop the second

I thought they had different properties (ref* vs. a smaller set of direct properties) and that the class differentiation was a didactic tool for the spec. There may be better ways to explain this. (ericP)

> ISSUE: Why two properties for rr:predicateObjectMap and rr:refPredicateObjectMap? Better to have just one, and spot the difference by looking at the (ref)objectMap contained within

I presume the same motivation and outcome as for the RefPredicateMap.

> ISSUE: rr:graphColumn, rr:graphIRI, rr:graphTemplate should be replaced by a rr:graphMap that works just like subjectMap, predicateMap, objectMap. This would simplify the spec, and it's very likely that those graphMaps can be re-used multiple times in a mapping, so shouldn't be more verbose either.

+1 (ericP)

> ISSUE: rr:termtype capitalization. rr:termType would be better

+1 (ericP)

> ISSUE: rr:subject and rr:object (in constant-valued term maps) should not be allowed to be blank nodes; just IRIs/literals; we say that blank nodes cannot be shared between graphs, so specifying a blank node in the mapping and expecting the *same* blank node to occur in the output data is a bit strange.

+0 deferring to your wisdom (ericP)

> ISSUE: I'd like to specify multiple predicates with a single predicateObjectMap, such as rdfs:label + skos:prefLabel. Currently cardinality is exactly 1. Can we loosen the upper bound?

+1 (ericP)

Received on Wednesday, 6 July 2011 12:21:22 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:00:25 UTC