Re: RDFa Core 1.1 feedback

On May 2, 2010, at 13:52 , Toby Inkster wrote:

> On Sat, 1 May 2010 22:59:27 -0400
> Gregg Kellogg <gregg@kellogg-assoc.com> wrote:
> 
>> *   Section 9 Note 3: Indicates that if a subject has more than one
>> triple with a predicate of rdfa:uri or rdfa:term, all triples
>> associated with the subject should be ignored (how I read it). What
>> if there are multiple values for rdfa:prefix? Does this imply
>> multiple prefix mappings, is is this also invalid?
> 
> Indeed, this section needs tightening up.
> 
> We need to consider what mappings are established in these three cases:
> 
> 	## Multiple URIs
> 	[] rdfa:term "Person" ; rdfa:uri <a> , <b> .
> 
> 	## Multiple terms
> 	[] rdfa:term "Person" , "Dude" ; rdfa:uri <a> .
> 
> 	## Multiple terms and URIs
> 	[] rdfa:term "Person" , "Dude" ; rdfa:uri <a> , <b> .
> 
> And the corresponding rdfa:prefix cases. Also:
> 
> 	## Both these in the same profile
> 	[] rdfa:term "Person" ; rdfa:uri <a> .
> 	[] rdfa:term "Person" ; rdfa:uri <b> .
> 
> Relying on the order that the mappings are specified might work for
> profiles written in RDFa, but what about profiles written in other
> profile serialisations? In many cases, RDFa parsers will hook into
> third-party parsers for those, and it's unlikely they'll be able to
> rely on triples being returned in any particular order.

Absolutely. We should not rely on any order. I am tempted to require a one-one mapping to all cases...

> 
> A good method of tightening up the definition of profiles might be for
> the spec to provide SPARQL SELECT queries that should be applied to
> profiles to yield the lists of mappings. (Implementations would not be
> required to implement SPARQL, but could use any technique to extract
> the lists of mappings which yielded the same results.)
> 

That can be the case. Note that the official vocabulary uses OWL statements that aim at describing the features. Adding SELECT somewhere as an information might also work, but may be too much...



> I don't actually see any harm in allowing terms to be expanded to
> multiple URIs. So, for example the term "title" could be mapped to
> both dcterms:title and rdfs:label. This would allow for profiles to be
> created that, say, simultaneously provided data in FOAF plus Google's
> vocab.

As far as I know, terms can also be used in @about and @resource. That would mean that we suddenly several resources in @about, ie, several subjects; that would really (over)complicate the processing steps I am afraid. I think we should avoid this at this moment...

Ivan


> 
> -- 
> Toby A Inkster
> <mailto:mail@tobyinkster.co.uk>
> <http://tobyinkster.co.uk>
> 
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf