Game Development Reference
In other studies, data mining is used to find
and to analyze data in the Web. The Web-based
data explosion has created a demand between
executives and technologist for methods to gather,
study and analyze the main information useful
for organizations. The emergence of Web Min-
ing identifies and manages the opportunities and
dangers in the web. In this way, these techniques
are used to collect and analyze Web-based data
to help increase sales or identify threats.
Software Agents are often used in BI. A Multi-
Agent System (MAS) is a system composed of
several agents that interact among themselves. The
interaction can be in the form of message pass-
ing, negotiating or changes in the environment.
Multi-Agent Systems shows a great potential
in advancing BI, solving complex problems in
a changing environment. In (Bobek and Perko,
2006) the authors build a model proposing three
fields in order to use a MAS combined with BI.
These fields are: intelligence data acquisition,
intelligent modeling and intelligent information
In other cases, the researchers use fuzzy
logic and evolutionary algorithms to predict the
consumer behavior in a business environment.
A solution for business intelligence has been de-
veloped by i imaginary, a company specializing in
knowledge management. This solution is called
iCLIP ( i (imaginary Client Profiler). In (Tettamanzi,
Carlesi, Pannese and Santalmasi 2007) the authors
describe a fuzzy logic approach to data mining
used by iCLIP, the MOLE engine. They present
the results on a customer prediction case.
own business simulator development kit, which
contains a business model description language
(BMDL) and a business model development
system (BMDS). BMDL is a designed language
that describes business models specifying the
relations and definitions among the ways to take
decisions, business variables and user interfaces.
BMDS generates executable codes for the server
program (Morikawa and Terano, 2005). A second
approach is to use Machine Learning methods.
Machine Learning is a branch of Artificial Intel-
ligence where systems improve their performance
with experience, and for instance, architectures
with machine-learning agents that substitute hu-
man players have been proposed.
Some Business simulation games use a mixture
of human and machine-learning agents. The learn-
ing agents can use a typical genetic-based learning
classifier system, XCS. In that study (Kobayashi
and Terano, 2003), the authors developed four
kinds of agents as alternatives to human players.
They implemented a random agent that makes its
decision using uniform random numbers; a reac-
tive agent whose decisions are based on the values
of several variables; agents that imitate human
behavior using the log information obtained in an
interaction between the simulator and a human; a
learning agent that uses a reinforcement learning
approach to acquire action policies.
Reinforcement learning (RL) allows decision-
making agents to learn from the reward obtained
from executed actions and, in this way, to find an
optimal behavior policy (Sutton and Barto, 1998).
In stochastic business games the players take ac-
tions in order to maximize their benefits. While
the game evolves, the players learn more about
the best strategy to follow. With this, RL can be
used to improve the behavior of the players in a
stochastic business game.
Machine Learning for
To implement a business simulator is a hard task
because it requires knowledge of computer sci-
ences; the game must be played again and again
to debug it, and many humans have to play it
until correct behaviors are obtained. In one ap-
proach, some researchers have developed their