[Modeling] Modeling an Agent Class- register your opinion

Dr. Hong Zhu hzhu@brookes.ac.uk
Thu, 19 Jun 2003 09:05:18 +0100

Dear Jim,

The choices of strategies is just a rough idea. Since UML 2.0 is not a
language based on traditional model of OO and it has extended OO concepts
quite a lot, adding another set of concepts (i.e. agents, agent classes,
etc.) into the language is quite hard, and adds to the complexity of the
language. I have difficulties to fit my meta-model into the system without
introducing a lot of things and redefining a large number of concepts.It
seems everything is already in the language, but they are slightly different
from what I really want.  The most important question is that:  by adding
new concepts of agents and agent classes (or caste), would we be able to
change the philosophy of modelling from object-orientation to
agent-orientation? For example, when I try to use the new AUML sequence
diagrams to model multi-agent systems, I have a feeling that I am still
thinking in a message driven approach. It is not quite an agent approach, in
which an agent senses the situation in the environment and takes appropriate
actions that other agents may sense its occurrence and take their actions or
ignor it as they like. Do you have the similar feeling?  Or have I missed
some thing?

The differences between the choises, IMO, is that: For choice 2, the outcome
would be a language that is upward compatible from UML to AUML in the sense
that any model in UML is a model of AUML. However, a model in AUML is not
nessarily a model of UML. For choice 1, we would first defind a core part of
UML which only supports traditional OO, and then extend it to include agent
concepts. The core UML will be compatible with standard UML in the sense
that a model in core UML is also a model of UML, but not the other way
round. The core UML would be the part of standard UMlL that does not include
non-traditional OO features. The consequence would be that a model in UML
maybe not a model of AUML, which, I guess, many people in this mailing list
don't like to see. However, by doing so, we might be easier to achieve the
change of modelling philosophy by shifting the focus of modelling from
object-orientation to agent-orientation. The above is just a vision. Sorry
for not having much details, but I hope it can be a starting point for the
discussion. What do you think?

Best regards,

----- Original Message ----- 
From: "James Odell" <email@jamesodell.com>
To: "ModelingTC" <modeling@fipa.org>
Sent: Thursday, June 19, 2003 3:16 AM
Subject: Re: [Modeling] Modeling an Agent Class- register your opinion

> On 6/16/03 6:34 AM, "Dr. Hong Zhu" indited:
> > Having said these about OO, the question is: Do we have a choice of
> > strategies to the design of AUML?
> > For example,
> > Choice 1: Come back to the simple and nice traditional model of
> > object-orientation and then add on agents into the model. Of course, we
> > not going to throw out the baby (i.e. the traditional OO) with the
> > bathwater.
> > Choice 2: Add on top of existing UML 2.0, which have already extended
> > tranditional OO model very much, with agent concepts.
> >
> > In my opinion, the choices between such strategies are vital to the
> > of the AUML and deserves a discussion in this mailing list.
> Yes, we have a choice of strategies to the design of AUML?  However, I am
> not sure I clearly understand the differences between Choice 1 and 2.
> 2.0 is not a "simple and nice traditional model of object-orientation."
> you wish to start with such a model, what would that look like? I know of
> standard model to use in this manner.  It would certainly not be extending
> UML.  Maybe that is your point?  Perhaps you could give us a little more
> detail on these choices.  I am a bit confused why these are "the choices."
> Regards,
> Jim
> _______________________________________________
> Modeling mailing list
> Modeling@www.fipa.org
> http://fipa.org/mailman/listinfo/modeling