[foaf-dev] Time to make the foaf classes relate to Dublin Core classes?

[Mikael Nilsson]

On ons, 2008-01-23 at 10:28 +0100, Danny Ayers wrote:
> >From the human definitions, it seems like Agent/Agent and
> AgentClass/Group are equivalents.
> 
> But it also seems like dcterms is a bit broken, having AgentClass as
> rdfs:Class and rdfs:subClassOf rdfs:Class. This definitely makes the
> vocab OWL Full, but I suspect it may also lead to
> unintended/undesirable inferences.

OWL Full is an issue (unavoidable?). But I fail to see the undesirable
inferences. Examples?

> 
> Agent and AgentClass are used elsewhere - e.g. audience range
> AgentClass, and note that :
> 
> dcterms:Agent
>     a    dcterms:AgentClass, s:Class .

This looks correct to me. In prose, it means that the class of all
Agents is an AgentClass. That sounds like a perfectly valid tautology to
me...

> 
> Seems like a simple statement like "Fred is doing the class DC101"
> would produce some very twisty RDF.

Namely?

> 
> I'm on the wrong machine right now, but it will be interesting to see
> what cwm and/or Pellet make of all this.

It would indeed, please let us know!

/Mikael

> 
> Cheers,
> Danny.
> 
-- 
<mikael at nilsson.name>

Plus ça change, plus c'est la même chose