view . spring 2009
7
Opposite page: Dr. Ziad Kobti is searching for answers.
What may have happened
to the
Ancient Pueblo Indians is one of the
enduring archaeological mysteries of
North America.
Now, however, a University of Windsor
scientist is using artificial intelligence to
help archaeologists solve the puzzle of
the group, commonly known as the
Anasazi, which disappeared more than
700 years ago.
Dr. Ziad Kobti, an assistant professor
in Computer Science, is part of a four-member team that received
almost $1.5 million from the U.S. National Science Foundation to
determine what may have happened to the Pueblos. “Archaeologists
have some solid ideas about where they might have gone, but they
don’t know for sure,” says Kobti.
Kobti is a part of the Village Ecodynamics Project, led by
researchers at Washington State University, and his share of the
funding is about $186,000. As the main programmer, he uses a
computer simulation technique, known as agent-based modelling,
to predict where prehistoric people might have situated their homes
depending on the natural and social environments in which they lived.
The Anasazi may have emerged around 1200 B.C. Native
populations grew rapidly between A.D. 600 and 1200 in the
Mesa Verde and Rio Grande regions of the American southwest,
located where the states of Utah, New Mexico, Arizona and
Colorado intersect.
The culture is best known for the stone and adobe dwellings
they built along cliff walls. The villages, referred to by Spanish
settlers as pueblos, were often only accessible by rope ladders
and rock climbing.
Then, for unknown reasons, the area was depopulated by
1280 A.D. Archaeologists have various theories, such as prolonged
periods of drought, global or regional climate change, disease,
the emergence of new social, religious or cultural influences,
raiding groups that created competition for resources, and even
cannibalism, though none of them have been verified.
Using archaeological data,
Kobti creates a virtual model of the
1,500-square-kilometre area as it may
have existed in 600 A.D. Each agent
represents a household with an average
of six family members. Those households
function as a unit and interact with
one another and the programmer can
introduce new influences, such as drought,
or new animals to hunt.
“The agent has autonomy,” says Kobti.
“It can think for itself without being told what to do. It can hunt,
plant crops, trade food and goods with other agents and make
decisions about where to live. It follows a strategy for survival and
you can test these strategies.”
Anyone expecting to see something akin to a video game,
however, might be disappointed. The model Kobti created does
not include actual visual representations of the agents, and the
experiments he conducts produce a series of diagrams and graphs
only a computer programmer could decipher.
By mimicking processes, such as population growth and
resource usage, the computer can track the migration of agents
and predict where the Native Americans may have moved. Those
predictions can be compared with existing archaeological data to
test their validity.
Kobti will conduct hundreds of such experiments. Each
experiment represents a time period of about 700 years, which
takes several days to complete on high performance systems.
The experiments are conducted on the SHARCNET
(Shared Hierarchical Academic Research Computing Network),
a consortium of academic institutions that share an infrastructure
network of high performance computers designed to accelerate
computational academic research and solve highly
complex problems.
The process, Kobti says, has universal potential applications.
“We can do the same thing to help solve historical and
archaeological mysteries right here in Essex County,” he says.
n
v
IN SEARCH
CIVILIZATION
of a lost
A UNIVERSITY OF WINDSOR
SCIENTIST IS USING ARTIFICIAL
INTELLIGENCE TO HELP
ARCHAEOLOGISTS SOLVE THE
PUZZLE OF THE GROUP, COMMONLY
KNOWN AS THE ANASAZI, WHICH
DISAPPEARED MORE THAN 700
YEARS AGO.
BY STEPHEN FIELDS
