Information Technology Reference
In-Depth Information
environment
agent A
agent B
m
0
A
(m
0
A
)
B
q
B
s
A
q
A
q
B
m
1
A
(m
1
A
)
B
q
A
s
B
m
2
B
(m
2
B
)
A
Fig. 3.9
Schema illustrating the process of negotiation of the common plan
The agent
B
compares both models it received from the agent
A
—the initial
(issue) model
B
and after evaluating
the models (changes) with the use of its objective function
q
B
, states that these
changes are not advantageous from its point of view. However, it goes further and it
finds that if it applies its strategy
s
B
m
0
)
B
and the target (destination) model
m
1
)
(
(
B
then the cumulative changes
in the environment are positive from its point of view (due to the evaluation with the
use of the objective function
q
B
—the comparison between the models
m
1
)
to the model
(
m
0
)
B
and
(
m
2
) and sends information about the application of the strategy
s
B
to the agent
A
B
and
m
2
).
The agent
A
receives information about the intentions of the agent
B
, but par-
ticularly it creates a model of suggested changes in the environment—the model
(
m
1
)
(
(and precisely the information about models
A
, which is evaluated with the use of the objective function
q
A
. If the changes
in the environment are advantageous to both agents, i.e. the evaluation with the use
of the objective function
q
A
m
2
)
m
0
,(
m
2
)
A
by the agent
A
and
q
B
m
0
)
B
m
2
)
by
the agent
B
give a satisfying result, then the agents realize their strategies. In effect,
the strategies
s
A
and
s
B
—a negotiated common plan (and goals)—are realized by
the agent
A
and the agent
B
(the strategy
s
A
by the agent
A
and the strategy
s
B
by
the agent
B
).
As a result of the negotiations, a plan of activity is created which will be realized
by a pair of agents and further by a group of agents within the multi-agent system.
It allows for cooperation of agents for the purpose of group problem solving.
(
)
((
,
3.6.2 The Process of Agent's Learning
The model of the M-agent may also be used for the description of the agent's process
of learning. In particular we may define two ways of realizing the process of learn-
ing, which seems very useful for improving the functionality of an agent in the
environment.