A great diversity of Smart Agents
"The agents' behavior is a result of its observations, intelligence and interactions
with the other agents. An agent is an autonomous and intelligent object that
disposes of its own mailbox in order to receive messages, those of its message
interpreter included. It has the capacity to refuse a message. You can affect
its goals and define what stabilizes and destabilizes it. In effect, in communicating
amongst themselves, the agents find the solution". While an object cannot execute
his methods, an agent has the competence that permits it to achieve its goal.
Marvin Minsky uses the metaphor of the tool and the collaborator to illustrate
the evolution of computer science techniques: the classic software are the tools
that the user guides directly, while the agents solve the problem by working
together. This is how people speak of "assistants" who treat the specific jobs
in the place of the user, that is to say that the research of information on
a network or a service. For example, Iris, the platform of the information
search of COGNITIX, presents an agent intelligence technique capable
of doing the intelligent documents search and the composition of services.
There is a great diversity of agent intelligence, and almost as many definitions
as societies that develop. The problem that actually presents itself, is the
dialog between the different agents. The operational solutions are generally
proprietary. The interest in the MindSuite Smart agents approach would be that different types
of agents could cooperate. ACL, A standard in Smart Agent communication between
agents, was actually developed by MindSuite Team. "With such an easy standard, all of the agents can communicate, regardless
of the language in which they were programmed (C, Cobol, Lisp, Fortran, Pascal, C++, etc) and regardless of the application
type".
12 May 1995,
Computer Magazine
 |