Computational Intelligence - August 2014 - 69

Notice that the belief space can be
viewed as a network of interacting
knowledge sources as shown in Figure 7.
Next, knowledge from the belief
space is allowed to influence the selection of individuals for the next generation of the population through the
influence() function. These two functions provide a feedback loop that allows
each of the two components to evolve
in parallel with the other. This works to
create a dual inheritance system in
which both components can evolve in
conjunction with each other. The Cultural Algorithm then repeats this process
for each generation until the pre-specified termination condition is met. In
this way, the population component and
the belief space interact with, and support each other, in a manner analogous
to the evolution of human culture.
There are a number of knowledge
categories which represent the behavioral patterns of individuals from the
population space that can be represented
in the belief space. Currently there are
five basic categories of knowledge that
can be stored in the belief space but
others are possible. These categories are
as follows [5]:
1) Situational knowledge: Situational
knowledge keeps track of the best
individuals who make the best
accumulative behaviors at each generation and stores them [6].
2) Domain knowledge: The domain
knowledge consists of the functional
relationships between various system objects and variables [7], [8].
This knowledge can be in the form
of equations, constraints, and other
symbolic methods used to represent
inter- and intra-object interaction.
3) Normative knowledge: Normative
knowledge provides standards or
norms for individual and group
behavior. It is often represented as a
set of intervals where each interval
is viewed as a promising range for
currently acceptable solutions for a
problem parameter [6].
4) History knowledge: History knowledge is used whenever search landscapes may change dynamically over
time in response to continuously

operating processes or discrete temporal spatial events. For example, it
is often used to keep information
about the sequences of environmental changes or social events over
time [7], [8].
5) Topographical knowledge: Topographical knowledge keeps a multidimensional grid or other logical
representation of the performance
landscape [7], [8]. This is often done
in terms of a geographical information system based upon a grid or
other network structures.
The actual implementation of each of
these knowledge sources in the Cultural
Algorithm used here will be discussed in
the subsequent sections.
D. Outline

In Section II the example site of Monte
Albán, is discussed. While the urban
center exhibited several occupational
phases we will focus on the first occupational phase, Ia, in order to identify
the early structural and functional urban
patterns that served to shape subsequent
phases. In Section III the data mining
activities used to initialize the belief
space in the Cultural Algorithm at each
of three different spatial scales are discussed. Data mining at the macro scale
produces the rules that determine what
terraces are occupied during that time
period. At the meso scale the functional
components of the city emerge: craft
production, residential, and elite residential. At the micro scale the basic
functional clusters of terraces are produced relative to these functional
classes. These will be the building blocks
that a population of planners in the
Cultural Algor ithm will combine
together to produce an abstract city
plan. Section IV then describes the
Cultural Algorithm configuration used
here. The agents in the Cultural Algorithm represent planners. These planners
will build regional models of the site
using the belief space knowledge as a
guide. Section V gives the results of the
evolutionary based planning process.
Section VI compares the results with a
model produced manually by a domain
expert and presents the conclusions.

Table 1 The early occupational phases
within the Valley of Oaxaca. Those
phases subsequent to the Rosario
phase are labeled by the
corresponding Monte Albán
occupation.
Period

aPProximaTe daTe

Tierras Largas

1400-1150 B.C.

san Jose

1150-850 B.C.

guadaLupe

850-700 B.C.

rosario

700-500 B.C.

MonTe aLBán ia

500-300 B.C.

MonTe aLBán ic

300-150/100 B.C.

MonTe aLBán ii

150/100 B.C.-a.d. 200

MonTe aLBán iiia

a.d. 200-500

MonTe aLBán iiib

a.d. 500-700/
750-1000

MonTe aLBán iV

a.d. 700/750

MonTe aLBán V

a.d. 1000-1521

II. The Site of Monte Albán

Given the urban models presented in the
previous section, to what extent does the
structural functional organization of this
early site relate to those of other modern
and ancient sites? What are the similarities
and the differences? Can these similarities
and differences be used to identify the
aspects of site formation here? In order to
answer these questions, we will look at a
particular city that emerged around 500
B.C., Monte Albán, located in the Oaxaca
valley of Mexico. Monte Albán is an
archaic urban center of over 300 hectares,
of which 70 hectares were densely occupied. It had an estimated 30,000 to
60,000 inhabitants in A.D. 600 [9].
The archaeological site of Monte
Albán is situated in Mexico's Valley of
Oaxaca, located in Central Mexico. Table
1 gives all of the relevant periods of social
evolution in the valley relative to the
emergence and occupation of Monte
Albán [9]. The Tierras Largas phase marks
the beginning of early village settlement in
the valley. The site of Monte Albán
emerged during Monte Albán Ia. The valley came under control of a state centered
at Monte Albán by Monte Albán II.
Monte Albán IIIa signaled the decline of
the state and its succession by a collection
of city-states localized in different parts of
the valley. The phases as described in the

august 2014 | IEEE ComputatIonal IntEllIgEnCE magazInE

Table of Contents for the Digital Edition of Computational Intelligence - August 2014