On Fri, 12 Nov 1999, David Bearman wrote:

> Andy, Rachel, Sigge, et.al,
> 
> I think the premise:
> > The word 'Role', as in 
> > DC.Creator.Role> 
> > does not refine Creator in the way that 'HasPart' as in
> > > DC.Relation.HasPart
> > > refines Relation.
> > > HasPart is a kind of Relation - Role is *not* a kind of Creator.At
> 03:37 PM 11/12/99 +0100, you wrote:
> 
> is wrong.

I agree with David. David, at the end of this mail I have a question for
you.

But will not discuss this further since we have a deliverable to complete
in DC-Agents. A majority in that group seems to be of the opinion that
what should go into that deliverable is a list of roles, and I agree that
that is a sensible thing to have before the end of this millenium. 

I do still have problems, though. One is that this we arbitrarily accept
the 'type' as a qualifier. That is, the one which we have traditionally
been using in the dotty syntax. DC.Creator.Organisation seem to be an
equally good refinement as DC.Creator.Illustrator. What we don't
cancatenate with dots have to be put somewhere... I can live with this
problem though, since it is a matter of syntax. I think. 

Nevertheless, I'm still sort of obsessed by the idea that there are sets
of objects called elements, and a yet to be discovered algebra that
operate on those objects, those that find this idea will have to bear with
me. Sorry. 

> If we correctly understand that Creator here refers to an Agent in a
> creative relation to the resource, the Role qualifies that relation. 
> Perhaps it is clearer with the DC.Contributor.Role since we understand
> contributor to be an agent in a relation of contributing and role as a
> qualifier of the kind of contribution. 

We have over and over again observed that elements, and not only values,
can be structured. Also, elements can be subject to fission and fusion
with preservation of semantics, i.e., DC.Creation can be split into

XX.Agent.Name
XX.Agent.Role = Creator

Having played around with this, it seems that fusions always involves
merging the semantics of a subelement into the semantics of a main
element. 

> The words we have adopted to describe types of agents are trying to
> denote specific types of relations (publish, create, contribute) to the
> resource.  In our resource-centric view of the world, agents exist only
> with respect to the actions that they take on resources....Role
> qualifies that action. 

David, in my current thinking, resource-centric and event-centric
descriptions can be transformed into each other, given that both adhere 
to the 1:1 principle.

Can you make up a counter-example?

Sigge




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%