Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com

Systems Engineering Procedia 00 (2011) 000000

Systems Engineering Procedia 2 (2011) 23 32

Systems
Engineering
Procedia
www.elsevier.com/locate/procedia

Engineering Modeling Unconventional Emergency Artificial

Society
Mingsheng Tang*, Xinjun Mao, Xueyan Tan, Huiping Zhou
Department of Computer Science and Technology, National University of Defense Technology, Changsha, China

Abstract
Lots of engineering modeling methods have been designed to investigate and analyze complex systems, such as artificial life,
cellular automata and artificial society. Artificial society is a bottom-up modeling method which is a new paradigm of simulating
human society from bottom up. A great amount of work has used the artificial society method. However, how to engineering
modeling artificial society with unconventional emergency is still worth to research. In this paper, we take into account the
particularity of unconventional emergency management to research artificial society modeling with the viewpoint of software
engineering. Then, we propose an unconventional emergency management-oriented artificial society model, including agent
model, environment agent model and emergency model. Furthermore, agents in the artificial society adopt the self-adaptive
approach binding mechanism, including role binding and emergency model binding. Then, the proposed model is illustrated by
an example of H1N1.

2011 Published by Elsevier B.V.

2011 Published by Elsevier Ltd. Selection and peer-review under responsibility of Desheng Dash Wu
Keyword: unconventional emergency; artificial society; agent; role; group; binding mechanism; engineering modeling

1. Introduce
As the rapid development of transportation, telecommunications and other industries have brought many benefits
to people's lives and social development, dissemination of information has became faster, interactions among people
have became more and more frequent, interactive areas are becoming more and more widely, and so forth. At the
beginning of this paper, let us look at a set of data about China:
In China, the total length of transportation was about 1.245 million km in 1980, and in 2006 approximately
reached 5.8187 million km, that increased 2.44-fold. Especially, in recent years China high-speed railway is
developing rapidly. Currently, high-speed railway operation mileage of China is longest in the world, and the
speed of China high-speed railway is also fastest in the world;
At the same time, the quantity of mobile phone users of China is more than 900 million, and China has become
the country with the largest mobile phone users in the world.
It also has brought some bad aspects: the speed and the scale of emergencies spreading in the society are
increasing, the frequency of occurrence of emergencies is also rising, and the local conventional emergencies may
evolve into large-scale unconventional emergencies. For example, H1N1 was outbreak in Mexico in 2009.
* Corresponding author. Tel.: +86-15116233906 .
E-mail address: tms110145@163.com.

2211-3819 2011 Published by Elsevier B.V.

doi:10.1016/j.sepro.2011.10.004

24

Mingsheng
et al.Engineering
/ Systems Engineering
Procedia
(2011) 23 32
Mingsheng
Tang Tang
/ Systems
Procedia 00
(2011) 2000000

Moreover, it continued to spread around the world, and eventually led to global impact.
Unconventional emergency is an emergency that has not adequate precursor, with significant features of the
complexity, potential hazards of secondary derivatives and the devastating, and is difficult to be dealt with the
conventional management approaches [1]. And unconventional emergency is the challenge and priority of the
current research of emergency management. In order to study the evolution and development of unconventional
emergencies in the society, researchers usually use some new methods, such as artificial life, cellular automata,
artificial society and so on [2]. In this paper, we focus on artificial society. The idea of artificial society method
proceeds as follow: firstly, through establishing the individual reality model (agent) in the society, agents must run
to comply with some rules in the system; Then, emergencies bring effect on the artificial society, and we observe the
emergence of agents behavior and properties in macro-scale to study the evolution of emergencies and the
appropriate emergency management. Therefore, the large-scale (tens of millions of agents) artificial society
modeling is one of the key technology in the domain of emergency management.
Artificial society establishes the complex virtual society which is similar to the complex real society with multiagents technology. In the recent, there are some multi-agent modeling and simulation platforms, including
Swarm(Santa Fe Institute), Repast(North and Macal), MASON(GMU) and Anylogic(XJTechnology) [8][9]which
are used for building large-scale multi-agent systems. Besides agents model, artificial society should establish the
social relations model. However, Swarm and MASON cant specify the social network of agents, and the model of
Repast and Anylogic is java model (so the work efficiency is limited). We make use of the idea of Model Drive
Architecture (MDA), and establish the artificial society model (agents, environment agents, social relations and so
forth) by graphical approach. Then, use the toolkit to transform the artificial society model to simulation model.
Lastly, experimenters deploy simulation models on the computing servers to experiment. For this, we propose an
artificial society model to build the artificial societies. In this paper, we introduce some knowledge of emergency
management and artificial society. And we aim at studying artificial society engineering modeling. Then, we
propose the artificial society model, including the formation of model elements, binding mechanisms and so forth.
2. Introduction of emergency management and artificial society
In order to solve the problem of complex social systems and the complexity of urban transport system, in 2004
[3]firstly proposed the social computing approach which consist of artificial societies for modeling, Computations
experiments for analysis, Computations experiments for analysis and Parallel execution for control, referred as ACP.
It is a new idea to solve issues of complex systems engineering modeling, analysis, calculation, control and
management, mainly including three steps. Firstly, establish the corresponding virtual complex system with the
method of artificial society engineering modeling. Then, design the computational experiments plan on the basis of
artificial society model with the method of computational experiments. Lastly, synchronously deduce the virtual
system and the physical system of the complex system, complementing each other, and manage the macro-scale
behaviours between them. Estimate the status of the evolution of the virtual system to adjust management
mechanisms and control methods in actual system. Thus, the study of artificial society engineering modeling is very
necessary, as its the foundation of the ACP method.

Mingsheng
al. / Systems
Engineering
Procedia
2 (2011)
23 32
Mingsheng Tang
Tang et
/ Systems
Engineering
Procedia
00 (2011)
000000

25

Fig1. ACP schematic diagram

2.1. Artificial society

Artificial society [4] [5] is a cross-discipline which combines sociology, computer science, MAS technology,
systems science and some other areas. Its a bottom-up modeling method, by creating the individual model in the
micro-scale and observing the emergence of behavior and attributes of the overall in macro scale. It may generate
from the nonlinear interactions between Agents. Artificial society is a general method for complex systems,
including multi-Agent technology, psychological aspect and complex social networks, and is used for simulating the
real societies or economic problems
Artificial society was initially used in economics, simulating some nonlinear economic phenomena such as stock
of non-linear Aurora. Nigel Gilbert and Rosaria Conte edited and published Artificial Societies The
Computer Simulation of Social Life in 1955, marking the artificial society as an independent new discipline, and
the study of artificial society was also carried out quickly.
In 1996, Epstein and Axtel edited and published Growing Artificial Societies - Social Science from the Bottom
up, constructing a classical model of artificial society Sugarscape [6]. Its a general model which is widely
applied to areas such as social sciences and economics. Two types of entities have been defined, environment and
agent. Environment is an area including a lot of resources, distributed according to some topology. Agent will
consume its resources to survive. Agent is the person in artificial society, and each Agent has its own internal
state and behavior rules. Agent needs to consume resources to survive, which should be searched in the
environment by Agent. According to the requirement of the research, the behavior of Agent will be changed by
changing environment regulations and rules. At the same time, explanation of some phenomenon or some new
conclusions will be obtained by observing emergence of behavior and properties of the artificial society with
simulation.
Unconventional emergency management-oriented artificial society model needs to create a city-scale artificial
society, including the environment agents, agents and so on. For its specificity, the behaviour, properties and rules of
a single agent must be as simple as possible, and it needs to be able to dynamically load emergency models during
the run-time to do emergency experiments.
2.2. Emergency Management
Emergency management is born out along with the emergencies. In the process of emergency prevention,
incident response and management disposal, government and public sector organizations take a series of response
and management control measures to safeguard social stability and reduce the impact of emergencies. Emergency
management involves prevention and emergency preparedness, detection and early warning, emergency response
and rescue, and the subsequent recovery and reconstruction, that is the process of Forecast - Warning - Response
Recovery. For unconventional emergency management needs to respond to natural disasters, accidents, disasters,
public health and other public emergencies that are with properties of sudden, evolution, extensive impact and so
forth.
Unconventional emergency management is the emergency management that takes into account unconventional
emergencies with significant features of the complexity, potential hazards of secondary derivatives, complexity, so
unconventional emergency management needs to have a predictive capability, early warning capability, which
requires artificial society model to describe real society accurately, and can well describe the social development
under the influence of unconventional emergencies.
2.3. A sample of unconventional emergency
Unconventional emergencies have significant features of the complexity, potential hazards of secondary
derivatives, complexity and so forth. We can understand what is unconventional emergency with a case. March
2009, H1N1 was outbreak in Mexico, and soon spread rapidly to more than 200 countries in the world. More than
tens of millions of people were infected, and it caused over 10,000 deaths and serious consequences. In the short
period of H1N1 outbreaks, H1N1 virus spread rapidly in the population, especially in schools, communities and

26

Mingsheng
et al.Engineering
/ Systems Engineering
Procedia
(2011) 23 32
Mingsheng
Tang Tang
/ Systems
Procedia 00
(2011) 2000000

other areas with dense population. So we can see the event is an unconventional emergency, with the characteristics
of suddenness, no foreboding, complexity and affecting widely.
3. Artificial society model
Artificial society model needs to establish the agent population, and needs some statistics to describe the
population agent as a whole, so we need a statistical model. We can use groups to characterize the social relations
and collective psychology of agents. To support multi-resolution model and simulation, we need environment agent
models. At the same time, in order to describe the emergencies, we need emergency models. Thus, artificial society
model of unconventional emergencies management-oriented contains Agent model, environment agent model and
emergency model, shown in Fig2. Compared to the general Multi-Agent System (MAS), the number of Agent in this
model is enormous (city-scale artificial society). Therefore, agents behaviour and rules need to be as simple as
possible.

Fig2. Artificial society model schematic diagram

3.1. Agent model

Agent Model is the model that is needed to construct agents. Its the basis of artificial society model, constituted
by statistical model, role set, action set, group and stochastic interaction model.
3.1.1 Statistical model
Statistical model describes the statistical properties of agents population in artificial society, such as total
population (Total number of agents), sex ratio, age distribution and so on, describing the agents population overall.
The total number of agents describes the population size of artificial society to be constructed. The role ratio
describes the proportion of various roles in the artificial societies. Role is mainly used to describe employment,
function and behavior of agents in the artificial society. Its the identity of agents in the artificial society. Sex ratio
denotes the sex ratio of agents in the artificial society. And age distribution table can count different ages or specific
ages as needed, describing the proportion of different ages of agents.
These basic models simply describe the composition of agents in the artificial society.
3.1.2 Action set
Just like human in real society, agent has his own behavior in artificial society. Action set defines the actions of
all agents, such as work, move and so forth. Action ID is the identity to distinguish different actions. The
prerequisite constrains the action execution, which is the condition that must be met before the action execution.

Mingsheng
al. / Systems
Engineering
Procedia
2 (2011)
23 32
Mingsheng Tang
Tang et
/ Systems
Engineering
Procedia
00 (2011)
000000

27

Execution process defines the specific content of the action. Execution results show the state of agent after execution
or the state of other elements to be changed. Agent description is the description of the behavior, which can be
described in natural language or other forms.
3.1.3 Role set
Role set defines the categories of roles in the artificial society. Role ID is the identity to distinguish each role.
Role name is the name of the role which specifies the occupation or function of an agent. A set of attributes will be
used to describe the properties of the different roles. For example, for the student role, there may be properties
named colleges, departments, grades and so forth. However, for a teacher role, there may be properties named
college, department, subject and so on. Roles can also be divided in different granularities based on the need of
simulation. For example, according to different grades, the student role can be divided into freshman role,
sophomore student role and so on. All these will be considered while modeling. Schedule defines a regular trip of
the role, which shows the possible actions of the role in each time period. In a period of time, a role may own
multiple choices of actions.
3.1.4 Group model
There are many differences between artificial society and the general Multi-Agent system (MAS), and one of
them is that in the artificial society agents own their social relations. Group model is used to describe a kind of
social relations between different agents, such as the dormitory relations, etc. Member table of the group describes
the group members. When the group composition relates to roles such as the relationship between teachers and
students, the model need to give out the related roles and the scale of the related roles. When the group composition
does not relate to roles such as friendship, the model will only give out the member size or scale. As the description
of the structure and organization of social relations of agents in the artificial society, the group will have the
frequency of activities, and in the group agents will have collective activities schedule together. And the activities
schedule will be described with time and activities. Interaction between agents will exist in the group. However, not
all of agents will interact with other agents in the group, and not every minute all agents will take part in the group.
Thus, the size or scale of the members to participate the group interaction is given by the scale of the interaction.
When taking part in group, the interaction probability is used to describe the possibility of agent interaction. For
some special groups, such as unconventional emergency committee, some group decision will be made for the
development of the emergency situation, thus affecting the development of the emergency in the artificial society.
At the same time, to describe the artificial society group psychology, such as the group psychology in public safety
incident, the group psychology could be specified by the group's action sequences, which will show the behavior of
the group. For example, in Xinjiang "7.5" incident, insurgents constituted a group to the produce violence
psychology through discussions, thus effecting on the society stability.

Fig3. (a)Group model; (b) Dormitory relationsocial relation

3.1.5 Stochastic interaction model

28

Mingsheng
et al.Engineering
/ Systems Engineering
Procedia
(2011) 23 32
Mingsheng
Tang Tang
/ Systems
Procedia 00
(2011) 2000000

The stochastic interaction model is used to describe the interaction between agents which happens outside the
groups. This interaction may be contact with the space and events, such as the spread of virus, or the interaction
between the two agents with social connection in some space. These interactions are stochastic, so we use the
probability of the stochastic interaction to describe the possibility of the interaction between agents.
3.2. Environment agent model
Environment agent model is mainly to describe some environment entity, such as classroom and library. The
environment agent model has its own function and actions, such as that it can either calculate the density of
population, or provide resource to the agent. The environment agent model is a necessary part of the artificial
society. Environment agent has a geographical scope, described by the height and the environment agent boundary
point set about longitude and latitude. The environment agent's resources is limited, therefore the environment agent
has a upper limit of population. Environment agent has his own activities, such as the library opens from 8:00am to
10:00pm. There is a affiliation between environment agents. For example, a didactical building has 50 classrooms,
classrooms belong to the didactical building, and didactical building contains an environment agent set (50
classrooms). With the dynamic evolution of the artificial society, agents activities are also dynamically changing in
the environment, and the environment agent records all active agents in the current environment with the population
agents set. So the environment agent can easily calculate the population density, spread of the disease, etc. Driving
by the unconventional emergency, environment agents will have some emergence behavior in the artificial society.
For example, calculation emergence, population emergence (agents may gather in the environment agent), etc.
3.3. Emergency model
Facing the artificial society model of the unconventional emergency management, emergency model is an
important component of artificial society model. Emergency will affect different agents behaviour, so different
emergencies will produce different behaviour. We use the emergency action set to describe the actions triggered by
the emergency, such as the H1N1 virus model can cause the action"self-test temperature". In artificial society, for
different roles the emergency effect is different. Emergencies mainly affect the agents activities schedule: hang
agents role activity schedule and make the emergency activities schedule as the agent activities schedule.
Emergencies conducing to changes of activities schedule is taking into account different roles, and the impact is
likely to be the same, so we use the role set to record these kinds of roles(with the same impact).
3.4. Self-adaptive mechanism binding mechanism
Agents need some self-adaptive methods and mechanisms. We use the self-adaptive mechanism - binding
mechanism in artificial model, including role binding [7] and emergency model binding mechanism. Role binding is
mainly used to simplify the functions, and agents behavior, properties and activities schedule related to roles should
be located in roles. If an agent plays a role, it will obtain all the properties, behavior and activities schedule of the
role at the same time. And agents can exit or enter a role to adapt to environmental change by dynamic role binding.
Emergency binding is mainly in response to emergencies, similar to the role binding, and agents will obtain
properties, actions and activities schedule of the emergency model. Then, the activities schedule of roles played by
the agent will be suspended to adapt to the impact of emergencies.
3.4.1 Role binding mechanism
Role is an abstract representation of agents behavior and function. Role binding mechanism is that give the role
to the agent and the agent has the properties, behavior and activities schedule of the role, etc. In artificial society, we
could bind multiple roles to an agent. And according to the simulation needs, one or more roles can be dynamically
bound to an agent. With the advance of a continuous evolution, the agent will enter one or more new roles and go
out of some roles. For example, after graduation some students will exit form the student role and may enter into the
teacher role.

Mingsheng
al. / Systems
Engineering
Procedia
2 (2011)
23 32
Mingsheng Tang
Tang et
/ Systems
Engineering
Procedia
00 (2011)
000000

Fig4. Role binding mechanism

3.4.2 Emergency binding mechanism

Emergency model binding mechanism will give some properties and changed schedule of the emergency model
to an agent, and the agent will suspend the activities schedule during normal circumstances. In the real society,
many emergencies may exist at the same time. For the artificial society, a variety of unconventional emergencies
may also act on the same agent. Therefore, we support that multiple emergencies models could dynamic bind on an
agent. After the elimination of an emergency, the emergency model will dynamically unbind from the agent, thus
the agent will exit the emergency model.

Fig5. Emergency binding mechanism

4. Case Study
Unconventional emergencies - H1N1 outbreak mentioned in section 2.3, we focus on a university campus to
establish its artificial society model. Firstly, give the statistical model of agent model, including the total population:
10,000 people, male and female sex ratio: 1:1, and the proportion of roles is shown in Table 1 and the age
distribution is shown in Table 2.
Table1. The proportion of roles
Roles

Teacher

30%

Student

50%

Service personnel

9%

Medical personnel

10%

Security guard

1%

Age

18-25

55%

25-35

15%

Table2. The age distribution

29

30

Mingsheng
et al.Engineering
/ Systems Engineering
Procedia
(2011) 23 32
Mingsheng
Tang Tang
/ Systems
Procedia 00
(2011) 2000000
35-60

20%

60-100

10%

Then, give roles set in artificial society, including student, teacher, service personnel, medical personnel and
security guard. We take the student role as an example to describe the student role. The student roles properties set
is including college, department, class, dormitory and son on, where its activities schedule can be shown in Table
3[15].
Table3. Activities schedule of the student role
Time slice

Actions (%)

00:00-06:00

Sleep

06:00-08:00

A: Breakfast (a%) B: Exercise (b%) C: Sleep (c%)

08:00-12:00

A: In class (a%) B: Breakfast (b%) C: Self-study (c%) D: Sleep (d%)

12:00-13:00

Lunch

13:00-14:00

Siesta

14:00-17:00

A: In class (a%) B: Self-study (b%)

17:00-19:00

A: Dinner (a%) B: Group activities (b%) C: Exercise (c%)

19:00-22:00

A: Dinner (a%) B: In class (b%) C: Group activities (c%) D: Self-study ( d%)

22:00-24:00

Sleep

Action set includes Dinner, Exercise, Work and so forth. The probability of stochastic interaction is 0.05. For
artificial society of the campus, there are many groups, such as friends group, student union group and dormitory
group, etc. For example, student union group contains group constraints (participants must be students and the
university has only one student group), group composition (200 people scale), frequency of weekly activities, and
activities schedule (every Monday evening from seven to nine), interaction probability of 0.7, and Action sequences
can express the collective psychology student group.
Then, establish environmental agent model, and we take classroom-A as an example. Location_ID of A
classroom is defined as 007, and its subordinate environment agent ID is 001 (Teaching Building B). Its
geographical range is expressed as {0, 4, {(121.12223, 27.13124), , (121.12225, 27.13128}} (height 1, height 2,
and a boundary points set of its latitude and longitude). Define carrying maximum as 200 people. And classroom A
is able to record the dynamic agents in the environment, and to calculate the population density of the environment
and the spread probability of disease, etc.
Finally, establish the emergency modelH1N1 model, and emergency ID is 001. Actions set of emergency
contains Self-test temperature, Medical care, Self-Isolation and so on. The response of different roles for H1N1 is
different. For example, if students and doctors are infected with the H1N1, their activities schedule will be different.
Students may continue their activities, and doctors may be self-isolation. In addition, give the additional properties
of H1N1. After affected by H1N1, agent should dynamically load these properties, including immunity, vaccination
conditions, symptoms, duration of symptoms, diagnostic methods used, the spread probability and so forth.
As a result, we have established an artificial society model. When an agent is infected with H1N1, the agent
should bind H1N1 model and suspend the activities schedule of the role played by the agent to make the activities
schedule of H1N1 model work. So agents obtain self-adaptive capacity.
5. Related work
Currently, successful international similar project is the BioWar project of United States [10] [11] [12]. BioWar
is a scalable city-scale simulation platform, and it could simultaneously simulate the impact of unknown infectious
diseases, natural disasters and bio-terrorist attacks (more events) to urban population. Biowar establishes the
artificial society through the establishment of agent population model, agent activities model, environment model
and disease model. For Biowar, there is no concept of the role, and the social relations are specified by relationship
set of the agent. Meanwhile, in Biowar the environment is just the time, weather and climate, and holiday cycle (not

Mingsheng Tang
al. / Systems
Engineering
Procedia
2 (2011)
23 32
Tang et
/ Systems
Engineering
Procedia
00 (2011)
000000

including environmental entities, such as the classroom). Paper [13] used multi-agent technology to study the
Epidemic News Spread Characteristics without the concept of roles. So agent's behaviour will not be different of
roles. Paper [14] used multi-agent to study dynamics of contagious disease spread, and mainly studied disease
spread model. For agent model in the artificial society, there is no concept of roles. Therefore, agents response to
disease has no difference. Meanwhile, there are no environment entities (without multi-resolution model), and
agents are given only the GIS where agents locate.
This proposed artificial society model gives a more complete artificial society model, including the agent model,
emergency model and environment agent model. Furthermore, we use the concept of roles that the different roles
response to emergencies is different. In a sense environmental agent model and agent model is isomorphic. So it can
support multi-resolution model and simulation. At the same time, we use the concept of group to describe social
relations between agents in the artificial society and the group psychology.
6. Conclusion and future work
In recent years, as the contact between international societies becomes increasingly close, the impact of
emergencies has become increasingly widespread. Lots of countries are studying on the emergency management,
using the method or idea of artificial society to model and simulate. In this paper, we take into account
unconventional emergency management to research artificial society engineering modeling, giving a general model
of artificial society, including agent model, environment agent model and emergency model to support multiresolution model, and we propose agents self-adaptive mechanism - binding mechanism. And the model could
describe unconventional emergency management -oriented artificial society completely. The next work contains: (1)
Abstract the artificial society model and propose the artificial society meta-model for unconventional emergency
management. (2) At the same time, study the artificial society modeling language and model mechanism. (3) With
the idea of MDA, research the automatic transforming method from artificial society model to simulation model and
develop the toolkit. (4) Propose a development methodology of artificial society for the unconventional emergency
management. (5) Lastly, use artificial society model, methodology, modeling language and toolkit for the
engineering of artificial society (tens of millions of agents) in case of emergencies such as H1N1, SARS and public
safety emergency.
7. Copyright
All authors must sign the Transfer of Copyright agreement before the article can be published. This transfer
agreement enables Elsevier to protect the copyrighted material for the authors, but does not relinquish the authors'
proprietary rights. The copyright transfer covers the exclusive rights to reproduce and distribute the article,
including reprints, photographic reproductions, microfilm or any other reproductions of similar nature and
translations. Authors are responsible for obtaining from the copyright holder permission to reproduce any figures for
which copyright exists.
Acknowledgements
The authors gratefully acknowledge the financial support from National Nature and Science Foundation of
China under Grant Nos. 9102403, and 61070034, Program for New Century Excellent Talents in University.
References
1. Han Zhiyong, Weng Wenguo, Yang Liexun, and Zhang Wei, Backgrounds, targets, and organization of the major research plan Study on
unconventional emergencies management. China Academic Journal Electronic Publishing House, 2009.
2. Zhang Jiang, and Li Xue-wei, Artificial SocietiesAgent Based Social Simulation. Systems Engineering, 2005.
3. WANG Fei-yue , Artificial Societies, Computational Experiments, and Parallel Systems: A Discussion on Computational Theory of
Complex Social-Economic Systems. Complex Systems and Complexity Science, 2004.

31

32

Mingsheng
et al.Engineering
/ Systems Engineering
Procedia
2 (2011) 23 32
Mingsheng
Tang /Tang
Systems
Procedia 00
(2011) 000000

4. Monbica Dascalu, Eduard Franti, Gheorghe Stefan, and Lucian, Artificial Societies: a new paradigm for complex systems modeling. 7th
WSEAS Int. Conf. on Arfificail Intelligence Gence, Know Engineering and Data Bases (AIKED'08), UK, 2008.
5. Reuben R. McDaniel, Jr. , and Dean J. Driebe, Complexity Science and Health Care Management. Advances in Health Care Management,
2001.
6. Epstein, JM. and Axtell R, Growing artificial societies - social science from the bottom up. Brooking Institution Press, Washington
D.C.,The MIT Press, Cambridge, Massachusetts &London, 1996.
7. Xinjun Mao, Lijun Shang, Hong Zhu and Ji Wang, The Adaptive Casteship Mechanism for Developing Multi-Agent Systems, International
Journal of Computer Applications in Technology (IJCAT), 2007.
8.Charles M. Macal, and Michael J. North, Agent-Based Modeling and Simulation. Proceedings of the 2009 Winter Simulation Conference,
USA, 2009.
9.Charles M. Macal, and Michael J. North, Toward Teaching Agent-Based Simulation. Proceedings of the 2010 Winter Simulation
Conference, USA, 2010.
10.Li-Chiou Chen, Kathleen M. Carley1, Boris Kaminsky, and Tiffany Tummino, Aligning Simulation Models of Biological Attacks.
Carnegie Mellon University, USA, 2004.
11.Carley, K.M., Altman, N., Kaminsky, B., Nave, D., and Yahja, and A., BioWar: A City-Scale Multi-Agent Network Model of Weaponized
Biological Attacks, Technical Report (CMU-ISRI-04-101). Pittsburgh, PA: CASOS, Carnegie Mellon University, 2003.
12.Dr. Kathleen Carley, Dr. Douglas Fridsma, MD, Dr. Elizabeth Casman, and Neal Altman, Scalable Multi-Agent Social and
Epidemiological Simulation of Bioterrorism. Events. Carnegie Mellon University, USA, 2003.
13.Xiaoguang Gong, and Renbin Xiao, Research on Multi-Agent Simulation of Epidemic News Spread Characteristics. Journal of Artificial
Societies and Social Simulation.No..31(2007).
14.Liliana Perez, and Suzana Dragicevic, An agent-based approach for modeling dynamics of contagious disease spread. International Journal
of Health Geographics, 2009.
15. Yuanzheng Ge, Wei Duan, Xiaogang Qiu, and Kedi Huang, Agent Based Modeling for H1N1 Influenza in Artificial Campus. The 2nd
IEEE International Conference on Emergency Management and Management Sciences (ICEMMS 2011), China, 2011.