Re: PROV-ISSUE-4: Coming to consensus on agent

If a relationship between an agent and a some process execution is required, then agent cannot be a type of thing and must be a role.

If it is only sufficient that an agent be ~capable~ of action then I think agent can be a type of thing - assuming there is agreement that a thing that is capable of action is always capable of action.

--Stephan

On Jun 28, 2011, at 5:01 AM, Paul Groth wrote:

> Hi All,
> 
> I believe that we were closer to reaching some consensus on agents. It would be good if we could get there by this weeks telecon as this would help in the process of consolidating concepts for the F2F meeting.
> 
> So far I believe the following are attributes that are agreed upon:
> 
> - An agent is active.
> 
> - Examples of agent are: people, organizations or computer agents (i.e. operating systems)
> 
> - There can be a relationship between agents and process executions
> 
> Discussion points:
> 
> - There was a debate around whether to make a human agent / computer agent distinction. Do we need to make this distinction?
> 
> - There was a debate about whether an agent should just be a status of a thing.
> 
> I think these should be able to be resolved quickly. I encourage people to look at the discussion on http://www.w3.org/2011/prov/wiki/ConceptAgent
> 
> I have added the following attempt at a summary definition.
> 
> PROPOSED: An agent is an active thing. It may be linked to a process execution, for example, by controlling it. Examples of agent include person, organization, and software agent.
> 
> Can you express your support for this definition or come up with an alternative by Thursday's telecon.
> 
> Thanks,
> Paul
> 
>

Received on Thursday, 30 June 2011 16:10:35 UTC