Information Technology Reference
In-Depth Information
concerning the interaction between agents (similar to those between objects) are
considered here and the solution to these problems will be suggested with the use of
a communication process and the operation of observation [31, 34, 60, 133].
The basic initial concept for further considerations is the idea that a particular
problem (task) may be solved not by one algorithm but by a group of cooperating
algorithms. In the beginning, the problem with cooperation of two (or more) coop-
erating algorithms appears. The model and then an attempt to define an agent will
be illustrated in the following steps:
•
Accepting as the starting point a general definition of an algorithm, used for solving
a specified task, we will consider the possibility of application (in the simplest case
two) of mutually cooperating algorithms in the realization of this task.
•
Further, we will consider the problem of algorithm decomposition which is too
complex (sophisticated) to be easily designed and realized; specifically we will
analyze how that kind of decomposition can be realized with the use of a few
cooperating simpler algorithms. The above- mentioned considerations on algo-
rithm decomposition will make it possible to define the relationship between the
cooperating algorithms.
•
The analysis of these relationships between cooperating algorithms leads to pro-
viding a more formal definition of the notion
autonomy
of a particular algorithm
towards other algorithms it cooperates with, as well as determining what conse-
quences arise from the lack of this autonomy.
Summing up, the above considerations lead us to the following conclusions:
•
Autonomy is not the characteristic (distinguishing) feature of the agent concept
because an object, or generally speaking a component algorithm, may also be
autonomous.
•
The distinguishing feature of the agent concept is the capability to observe its
surrounding environment, including other agents operating in a given environment,
which makes that the agent acquires the property of autonomy in relation to other
agents.
The analysis of these properties allows for decomposition of a particular algorithm
into component algorithms such as objects or agents. By comparing the property of
autonomy and the concept of encapsulation (in the sense of the object approach) it
will be possible to define an object and an agent, as well as their basic distinguishing
features.
2.2 The Model of an Algorithm and Problems
with Its Decomposition
Let us consider the following definition of an algorithm
Al
g
[149, 180]:
Al
g
=
(
,
),
U
F
(2.1)
