- From: Eric Prud'hommeaux <eric@w3.org>
- Date: Sat, 12 Nov 2011 18:06:02 -0500
- To: Richard Cyganiak <richard@cyganiak.de>
- Cc: Ivan Herman <ivan@w3.org>, Steve Harris <steve.harris@garlik.com>, Ian Davis <ian.davis@talis.com>, W3C RDF WG <public-rdf-wg@w3.org>
* Richard Cyganiak <richard@cyganiak.de> [2011-11-09 21:23+0000] > On 9 Nov 2011, at 20:26, Eric Prud'hommeaux wrote: > >> IMO W3C should: > >> > >> 1. Assign short URIs in a *single* namespace for all RDF and RDFS (and possibly OWL and XSD) concepts *now* > >> 2. Leave it up to individual WGs to adopt those short URIs at their own leisure > >> 3. Leave it up to implementers to add support for them already > >> 4. The RDF WG should *not* do anything about them in RDF 1.1, but perhaps in RDF 2.0 > > > > Every evolution path I see from this leads to either fragmentation or unrealistic implementation demands. The options I see are: > > a. do nothing. > > b. gradual introduction of redundant short terms, followed by gradual redaction of longer names. -- good bye cardinality > > c. international change-over day. -- all the air traffic control and clinical support systems relying on RDF will crash that day. > > d. stake out short syntactic forms for use in turtle in SPARQL but leave the denotations the same. > > Consider the evolution path for unifying "foo" and "foo"^^xsd:string: Both exist side by side in the wild and query authors have to deal with that. Increasingly, we will have systems that know that both are equivalent, e.g., those with D-entailment support and those with RDF 1.1 support. Both forms will continue to be deployed in the wild forever but which one to use will be merely a matter of taste. > > It would be the same with <http://n.w3.org/rdf/type> and <http://www.w3.org/1999/02/22-rdf-syntax-ns#type>. Just replace “D-entailment” with “owl:equivalentProperty entailment” and “RDF 1.1” with “RDF 2.0”. It was my impression that we decided that, for RDF 2012, "foo" is a shorthand for "foo"^^xsd:string. That is Data: <x> <p> "foo", "foo"^^xsd:string . Query: SELECT ?s { ?s <p> ?o } Results (RDF 2004): ?s <x> <x> Results (RDF 2012): ?s <x> and that we justified this by saying that no one really ever wanted them to be different in the first place. Following this scenario, we could say that <http://n.w3.org/rdf/type> and <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> are the same node but then what is STR(<http://www.w3.org/1999/02/22-rdf-syntax-ns#type>)? (I reallize that I betray my bias by casting everything in terms of query, but I think that SPARQL is the most used and useful tool in the SemWeb toolbox.) If I misunderstand, we have instead decided that D-entailment gives us two nodes, Data: <x> <p> "foo" . Query: SELECT ?s { ?s <p> ?o } Results (RDF + D-entailment 2012): ?s <x> <x> then I'm skeptical that we've simplified anything. (But I'll go along with it if SELECT (SUM(?salary) AS ?checkAmount) { ?who foaf:given_name "eric" ; mit:salary ?salary } doubles my salary.) > Best, > Richard -- -ericP
Received on Saturday, 12 November 2011 23:06:33 UTC