U. Brinkschulte, T. Givargis, and S. Russo (Eds.): SEUS 2008, LNCS 5287, pp. 1–8, 2008.
© IFIP International Federation for Information Processing 2008
Using UML 2.1 to Model Multi-agent Systems
Darshan S. Dillon, Tharam S. Dillon, and Elizabeth Chang
Digital Ecosystems and Business Intelligence Institute,
Curtin University of Technology Perth, Australia
{Darshan.Dillon,Tharam.Dillon,
Elizabeth.Chang}@cbs.curtin.edu.au
Abstract. The use of UML 2.1 to model a broad range of systems is evident
from the variety of UML diagrams in academia and in the marketplace. One
class of systems currently gaining popularity are Multi-Agent Systems. There
are efforts underway to use UML to model these systems and these efforts are
both productive and form the basis for both a methodology and a notation for
systems of this type.
1 Introduction
In this paper we first introduce what an Agent is, the key characteristics of an Agent,
the scope of this paper in terms of what we model in Multi-Agent Systems, and
finally future directions.
2 What Is an Agent ?
In order to define what an agent is we should first consider a definition from the
literature.
An agent is a computer system that is situated in some environment, and that is ca-
pable of autonomous action in this environment in order to meet its design
objectives.
1
From this definition a number of points are clear. Firstly, the location of the com-
puter program is important. This is so because the program can migrate from one
machine to another. This is not the usual pattern of behaviour for computer programs.
They usually are installed, configured and run on a particular machine. They do not
travel, as such. Secondly, the computer program is capable of acting automously,
which means it is not dependant on any other program. This goes together with the
fact that agents are mobile. They can be launched by a user on a particular machine,
and travel, severing their connection with the user and concentrating their state related
information within themselves. Thirdly, the computer program is goal-driven and can
choose to act in a way that satisfies it’s design objectives. Most computer programs
are data-driven, reacting to inputs.
Finally, agents play an important role in embedded and ubiqitious computer sys-
tems. They are particularly important in goal-oriented or mission-oriented environ-
ments. The modeling and design of agents is an important first step for the building of
agent- based systems.