Re: Organization types predicates vs classes

On Tue, Jun 8, 2010 at 12:21 PM, William Waites <william.waites@okfn.org> wrote:
> On 10-06-08 04:27, Todd Vincent wrote:
>>
>> By adding "OrganizationType" to the Organization data model, you provide
>> the ability to modify the type of organization and can then represent
>> both (legal) entities and (legally unrecognized) organizations.
>
> :foo rdf:type SomeKindOfOrganisation .
>
> vs.
>
> :foo org:organisationType SomeKindOfOrganisation .
>
> I don't really see the need for an extra predicate
> with almost identical semantics to rdf:type. There
> is nothing stopping a subject from having more than
> one type.

Yes, exactly. The schema guarantees things will have multiple types.
The art is to know when to bother mentioning each type. Saying things
are an rdfs:Resource is rarely interesting. Saying they're a
foaf:Agent is also pretty bland and uninformative. The mid-level
classes around Organization are generally more interesting, and folk
using local / community-extended classes (foo:CultLikeOrganization
bar:SomePreciseSubClassOrg etc) probably ought to mention mid-level
classes too. Some day we'll get support for these distinctions from
the big RDF aggregators and from analysis of code, SPARQL queries etc,
so we know which terms are most likely to be understood.

BTW the syntax of RDFa (compared to RDF/XML) makes it easy and much
less ugly to mention extra types and relations. Mentioning a second
relationship in original syntax of RDF/XML is particularly verbose. In
RDFa we have space-separated lists of qualified names, which
significantly reduces the cost of mixing general (widely understood)
classes with precise (but more obscure) community extensions. This is
a pretty good thing :)

cheers,

Dan

Received on Tuesday, 8 June 2010 10:57:21 UTC