Re: prov-dm: draft for the new 'hadPlan' relation

On Wed, Dec 14, 2011 at 14:45, Luc Moreau <L.Moreau@ecs.soton.ac.uk> wrote:
> I am in agreement with what you said.  That's a concern we also had.
> But it is a natural consequence of defining an agent as any entity being
> associated with an activity.

Yes, it falls out of "an entity that can be assigned some degree of
responsibility for an activity taking place".

But as you say, this could even just be an activity which was
triggered due to an entity appearing.  Some supermarket tills have
sensors so that the conveyor belt automatically continues forward
until there is an item in front of the cash register. If I put a food
item on the belt, I am not an agent triggering the belt moving
forward, because it goes forward even when empty. However the belt
will stop when the food item has been moved to the end.

So (if we ignore the sensor for now - I consider that low-level) by
the current proposal that food-entity can be said to be triggering the
belt-movement activity to stop. That means it is an agent. However the
food item has no "responsibility" to talk of as such, and have not
done any decisions beyond staying on the belt. It was the conveyor
belt's responsibility to stop.


> There, a request 'starts' the activity, and hence is also an agent.
> We may want to revise the notion of agent, and have it defined standalone,
> independently of wasAssociatedWith.  Then, wasAssciatedWith could be
> associated with entities and agents.

You are right, it seems the wasAssociatedWith-agent is much weaker
than the usual classification of "agent" - and could make it harder to
determine if an entity was "used" or otherwise associated with an
activity in an agent-like pattern.

If we make "agent" be stronger, we can make wasAssociatedWith weaker
so it only relates entities to agents. This would allow an activity to
be associated with entities which were neither used or agents in a
strict sense, for instance a building's air conditioning system
involvement in a chemistry experiment, or a tape recorder present (but
not turned on) during a police interview.


> I think we had a hard fought battle on agents at F2F1. I will not change the
> text until we have a new vote on agents.

Agreed, tread carefully. :)


-- 
Stian Soiland-Reyes, myGrid team
School of Computer Science
The University of Manchester

Received on Wednesday, 14 December 2011 15:15:05 UTC