Re: what to do with dc:creator in Annotea?

Marja-Riitta Koivunen <marja@w3.org> wrote:
>>
>> I'm not sure where this needs to be corrected. Could we have more structure
>> in dc:creator so that the applications could use it too or just ask the
>> server to put the two names together to a string?
>> 
>> How are other people using this?

Aaron Swartz replied:
> 
> Many people believe that dc:creator should point to a new RDF resource, with
> properties coming off of that, as opposed to just a text string. Something
> like:
> 
> <#annotation> dc:creator :X10101 .
> 
> :X10101 
>     dc:title "Aaron Swartz"
>     foaf:given "Aaron" ;
>     foaf:family "Swartz" ;
>     foaf:mbox <mailto:me@aaronsw.com> .
> 
> I'd think this is perfectly acceptable, because it contains the structured
> information, as well as the dc:title value for simpler processors that just
> want a string.

I basically agree with this, but I'd suggest doing it as a non-binary
relation (RDF M&S section 7.3) and using rdf:value instead of (or in
addition to) dc:title.  The RDF processor might not know that the property
dc:title is more appropriate than the others as a default literal value.
It does know that rdf:value is special.

As an aside, two vocabularies that lend themselves to describing structured
personal names are vCard and X.500.  Perhaps the following:

<#annotation>
  dc:creator     :X10101 .

:X10101
  rdf:value      "Aaron Swartz"
  x500:cn        "Aaron Swartz"
  x500:givenName "Aaron"
  x500:sn        "Swartz"
  x500:mail      <mailto:me@aaronsw.com> .

Ideally, you'd infer the rdf:value statement from the x500:cn (or dc:title)
statement rather than duplicating the literal "Aaron Swartz" in the source
statements, but expressing the inference rule would have to be done outside
of RDF.

-- 
PGP pub key: http://staff.pisoftware.com/raboczi/pgp-pub-key.asc
Fingerprint: 5DOC 23A5 D5C7 FFD1 030F 62B0 D168 C77A EB25 A3A9