[Modeling] Modeling an Agent Class- register your opinion

Dr. Hong Zhu Dr. Hong Zhu" <hong.zhu1@BTinternet.com
Sat, 21 Jun 2003 20:34:51 +0100

Hi, Jim,

> Since that was "so long ago," I have nothing in machine-readable form.
> of the prominent researchers that I easily recall is Brian
> <brian@it.uts.edu.au>.  He methodology actually proposed role diagrams.
> Perhaps, if you will contact him.  He is now fairly active in the agent
> field.

Thanks for the information.

> For the record, I was one of his detractors.  Even then, I was still
> challenging people to formally define the difference between class and
> -- because I could see no difference.  For me now, there is a
difference --
> but it is small.  As I mention earlier, for me a role is a class (or
> AgentClass, if you wish) whose instances interact.  Therefore, Employee is
> role because it is a class whose instances interact with instances of
> (or the same) role.

I would like to distinguish class from role as follows. Class is a language
facility that has well-defined structure and senamtics, while role is an
abstract concept that different people may have different understanding.
However, the concept of roles helps the analysis and design of systems. It
can be implemented by classes. Therefore, in my opinion, there is no need to
include a concept of role in the meta-model.


> -Jim
> On 6/20/03 9:27 AM, "Dr. Hong Zhu" indited:
> > It is nice to see that we all agree that role modelling is a part of
> > agent-oeirntation methodology. Could you please tell us more about role
> > modelling in the mid-80s, or if there is any differences from what
people is
> > doing with role modelling, now?
> _______________________________________________
> Modeling mailing list
> Modeling@www.fipa.org
> http://fipa.org/mailman/listinfo/modeling