Computational Network Science An Algorithmic Approach 1st Edition Hexmoor All Chapter Instant Download
Computational Network Science An Algorithmic Approach 1st Edition Hexmoor All Chapter Instant Download
com
https://textbookfull.com/product/computational-
network-science-an-algorithmic-approach-1st-
edition-hexmoor/
https://textbookfull.com/product/fault-tolerant-message-passing-
distributed-systems-an-algorithmic-approach-michel-raynal/
textbookfull.com
https://textbookfull.com/product/intelligent-video-surveillance-
systems-an-algorithmic-approach-first-edition-maheshkumar-h-kolekar/
textbookfull.com
https://textbookfull.com/product/an-introduction-to-computational-
science-allen-holder/
textbookfull.com
https://textbookfull.com/product/guide-us-home-1st-edition-jesse-j-
thoma-cf-frizzell/
textbookfull.com
The Laboratory Nonhuman Primate, Second Edition Jeffrey D.
Fortman
https://textbookfull.com/product/the-laboratory-nonhuman-primate-
second-edition-jeffrey-d-fortman/
textbookfull.com
https://textbookfull.com/product/the-macroeconomics-of-corruption-
governance-and-growth-1st-edition-maksym-ivanyna/
textbookfull.com
https://textbookfull.com/product/innovations-in-manufacturing-for-
sustainability-kapil-gupta/
textbookfull.com
https://textbookfull.com/product/regular-graphs-a-spectral-
approach-1st-edition-zoran-stanic-2/
textbookfull.com
Guardians of language : twenty voices through history 1st
Edition Coulmas
https://textbookfull.com/product/guardians-of-language-twenty-voices-
through-history-1st-edition-coulmas/
textbookfull.com
Computational Network Science
An Algorithmic Approach
Henry Hexmoor
Notices
Knowledge and best practice in this field are constantly changing. As new research and experience
broaden our understanding, changes in research methods, professional practices, or medical treatment
may become necessary.
Practitioners and researchers must always rely on their own experience and knowledge in evaluating
and using any information, methods, compounds, or experiments described herein. In using such
information or methods they should be mindful of their own safety and the safety of others, including
parties for whom they have a professional responsibility.
To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors, assume
any liability for any injury and/or damage to persons or property as a matter of products liability,
negligence or otherwise, or from any use or operation of any methods, products, instructions, or
ideas contained in the material herein.
ISBN: 978-0-12-800891-1
The days of the need for gurus and extensive libraries are behind us. The
Internet provides ready and rapid access to knowledge for all. This book
offers necessary and sufficient descriptions of salient knowledge that
have been tested in traditional classrooms. The book weaves founda-
tions together from disparate disciplines including mathematical sociol-
ogy, economics, game theory, political science, and biological networks.
Network science is a new discipline that explores phenomena com-
mon to connected populations across the natural and man-made world.
From animals to commodity trades, networks provide relationships
among individuals and groups. Analysis and leveraging connections
provide insights and tools for persuasion. Studies in this area have large-
ly focused on opinion attributes. The impetus for this book is a need to
examine computational processes for automating tedious analyses and
usage of network information for online migration. Once online, net-
work awareness will contribute to improved public safety and superior
services for all.
A collection of foundational notions for economic and social net-
works is available in Jackson (2008). A mathematical treatment of
generic networks is present in Easly and Kleinberg (2010). A comple-
mentary gap filled by this book is an algorithmic approach. I provide a
fast-paced introduction to the state of the art in network science. Refer-
ences are offered to seminal and contemporary developments. The book
uses mathematical cogency and contemporary computational insights.
It also calls to arm further research on open problems.
The reader will find a broad treatment of network science and review
of key recent phenomena. Senior undergraduates and professional peo-
ple in computational disciplines will find sufficient methodologies and
processes for implementation and experimentation. This book can also
be used as a teaching material for courses on social media and network
analysis, computational social networks, and network theory and ap-
plications. Our coverage of social network analysis is limited and details
are available in Golbeck (2013) and Borgatti et al. (2013).
x Preface
REFERENCES
Borgatti, S., Everett, M., Johnson, J., 2013. Analyzing Social Networks. SAGE Publications.
Easly, D., Kleinberg, J., 2010. Networks, Crowds, and Markets. Cambridge University Press.
Golbeck, J., 2013. Analyzing the Social Web. Morgan Kaufmann Publications.
Jackson, M., 2008. Social and Economic Networks. Princeton University Press.
CHAPTER 1
Ubiquity of Networks
1.1 INTRODUCTION
Broadly speaking, a network is a collection of individuals (i.e., nodes)
where there are implicit or explicit relationships among individuals in
a group. The relationships may be strictly physical as in some sort of
physical formation (e.g., pixels of a digital image or cars on the road),
or they may be conceptual such as friendship or some similarity among
pairs or within a pair. In an implicit network, individuals are unaware
of their relationships, whereas in an explicit network, individuals are
familiar with at least their local neighbors. In certain implicit networks
called affinity networks, there is a potential for explicit connections from
relationships that account for projected connection such as homophilly
(i.e., similarity) (McPherson et al., 2001). Biological networks capture
relationships among biological organisms. For instance, the human brain
neurons form a large network called a connectome (Seung, 2012). An ant
society is an example of a large biological network (Moffett, 2010). There
are many examples of small-scale animal networks, including preda-
tors and their prey, plant diseases, and bird migration. Human crowds
and network organizations (e.g., government or state agencies, honey
grids in bee colonies) are other examples of natural networks. Modern
anonymous human networks have capacities for crowd solving problems
(Nielsen, 2012), where a group of independently minded individuals pos-
sess a collective wisdom that is available to singletons (Reingold, 2000).
Social and political networks model human relationships, where social
and political relations are paramount. Economic networks are models
of parties related to economic relationships such as those among buyers
(and consumers), sellers (and producers), and intermediaries (i.e., trad-
ers and brokers) (Jackson, 2003). Beyond natural networks, there are
myriads of synthetic networks. The grid of a photograph is an example
of synthetic networks. Nanonetworks are attempts to network nanoma-
chines for emerging nanoscale applications (Jornet and Pierobon, 2011).
2 Computational Network Science: An Algorithmic Approach
(e.g., trust). Relational networks are inert and merely reflect juxtaposi-
tion of nodes. All CENs are exchange networks.
Once a network emerges, we can explore interactions within the net-
work. Strategic interactions involve reasoning and deciding over selec-
tion of strategies. They can be modeled with game theory that will be
our main focus in Chapter 3.
Network theory is a set of algorithms that codifies relationships
among network topology and outcomes, which are meaningful to net-
work inhabitants. There is a movement afoot that codifies network phe-
nomena under the term network science. These phenomena and salient
algorithms will be discussed throughout this book.
An Online Social Networking Services (OSNS) creates synthetic net-
works among people. The salient incentive for using an OSNS is to gain
social authority (i.e., legitimacy), which is a form of social power and
not generally a measure of vanity. Social authority in social networks
is with respect to a group and with respect to specific topics. Therefore,
social authority is a relative measure and not an absolute quantity. In
Section 1.2, we review a few popular OSNSs from a rapidly growing list
(Khare, 2012). Since they provide platforms to create, to share, and/or
4 Computational Network Science: An Algorithmic Approach
when a family stays organized about their travel itinerary, there are dis-
parate opinions. In the large network, a large social project, such as a
protest, can be planned. Twitter can be used to work semi-anonymously
with others. Twitter’s hashtag (i.e., #) is a feature for labeling a topic. Any-
one may introduce or reuse a hashtag to attract attention. For example,
#flight1549 added to a tweet labels the tweet to be about “flight1549.”
This hashtag labeling facilitates search related to specific topics. Individ-
uals who use specific hashtags form an implicit network in the context
of their hashtags. This feature has been used for commercial marketing
and anonymous coordination over social actions. The range of poten-
tial uses for hashtags is enormous, and they have been adopted by other
OSNSs such as Facebook. On the one hand, Twitter can be used for so-
cial organizations of crime or dissent. On the other hand, it can be used
to predict and mitigate violations of law enforcement. Since Twitter pro-
vides democratization of opinion sharing and equal access for dissemi-
nation, it is seen as a social equalizer and as such it might be feared by
repressive systems (e.g., government regimes). Twitter’s social authority
is composed of three components: (1) the retweet rate of users’ last few
hundred tweets, (2) the recentness of those tweets, and (3) a retweet-
based model trained on users’ profile data. Tagging someone shows
the Twitter id to more people, whereas direct messaging someone just
puts spam in their inbox, which is generally undesirable. Websites, such
as Klout.com, gauge the influence you have by monitoring things, for
example, how active you are and how much you have been tagged on
Twitter. Twitter’s lists are a way to organize others into groups. When
you click on a list, you will retrieve a stream of tweets from all the users
included in that group. As a rule of thumb, if you want to develop re-
lationships on Twitter, you should read other tweets, retweet good con-
tents, tweet good contents, and stay on top of keywords and interests
that you follow. The same advice applies if you want to get retweeted.
Linkedin is an OSNS that provides an online forum for professional
identity management. The main tool for Linkedin’s connections is to
link people, who would like to support one another (i.e., connections).
Linkedin allows people to conduct a weak form of endorsement in re-
gards to specific skills. This creates directional links from endorsers to
endorsees. Linkedin allows a stronger directional endorsement through
recommendations. Endorsed individuals’ profiles gain social authority
via Linkedin’s endorsements and recommendations. Of course, the
6 Computational Network Science: An Algorithmic Approach
1.4.1 Random Networks
G(n, p) is a random graph model with n nodes where the probability of a
pair of nodes in it being linked is denoted by p (Erdős and Rényi, 1959).
When p is small, the network is sparsely connected. When p is close to
1/n, the network appears fully connected. When p is almost 1.0, the con-
nectivity among nodes is very high and the network is said to be a giant
component. The spread of node degrees for a random graph model (i.e.,
degree distribution) appears binomial in shape. A closely related model
is the random geometric graph G(n, r), where there are n nodes and the
distance between a pair of nodes in the graph is less than or equal to r
(Penrose, 2003). Contrary to mathematical models, real-world networks
exhibit a degree distribution that is unevenly distributed. In the power-
law distribution, the probability that a node has a degree distribution k
(i.e., the number of connected neighbors) is determined by P(k) ≈ kg, where
parameter g is typically constrained between 2 and 3, that is, 2 ≤ g ≤ 3.
Uneven distribution stems from preferential attachment, where the prob-
ability that a new node will attach to a node i is degree( i ) / ∑ j degree ( j ) .
A node degree refers to the node’s number of neighbors. Preferential
attachment is commonly found in nature as well as man-made networks
such as an economic network (Gabaix, 2009). Random networks are
mathematically the most well-studied and well-understood models.
1.4.2 Scale-Free Networks
There is a model based on preferential attachment described by Bara-
basi and Albert (1999). In this model, a new node is created at each
time step and connected to existing nodes according to the “prefer-
ential attachment” principle. At a given time step, the probability p
of creating an edge between an existing node u and the new node is
p = [(degree( u ) + 1) / (| E | + | V |)] , where V is a set of nodes and E is the
set of edges between nodes. The algorithm starts with some parameters
such as the number of steps that the algorithm will iterate, the num-
ber of nodes that the graph should start with, and the number of edges
that should be attached from the new node to preexisting nodes at each
time step. The Barabasi model of network formation produces a scale-
free network, a network where the node degree distribution follows a
power-law principle. Scale-free networks produce small number of com-
ponents, small-diameter, heavy-tailed distribution, and low clustering.
8 Computational Network Science: An Algorithmic Approach
Many types of data studied in the physical and social sciences can be
approximated with a Zipf distribution (Li, 1992), which is one of the
families of discrete power-law probability distributions. An implication
of the Zipf law is that the most frequent word will occur approximately
twice as often as the second most frequent word, which occurs twice as
often as the fourth most frequent word, etc.
Unlike the growth model of Barabasi, Epstein and Wang’s (2002)
steady-state model uses a rewiring scheme that results in power-law
distribution. This model evolves an initial graph according to Markov
process, while maintaining constant size and density.
Epstein and Wang’s algorithm has two major steps: (1) initial-
ize a sparse graph and (2) edit Markov edges. To generate the sparse
graph G, they randomly add an edge between vertices with probability
2 m / [ n ×( n − 1)] , where m is the number of edges added and n is the
number of vertices. If the number of edges in G is still less than m, they
start adding edges with a probability of 0.5 until the graph G has m
edges. The second step is to reiterate the algorithm in Figure 1.2 r times
on G, where r is a parametric value.
1.4.3 Trade-Off Model
A trade-off-based model of network formation is the highly optimized
tolerance (HOT) class of models. In a simple model, nodes are allowed
to reason about their connections to other nodes. A node i’s connection
cost to a node j is denoted by cij. The node i will consider centrality of
potential nodes j for attachments denoted by cenj. The node i will con-
sider nodes that minimize the value of a × cij + cenj, where a is a positive
weighing factor dependent on the network size n.
The Transit(·) function specifies the flow rate among nodes of the net-
work. If we could specify the maximum flow between all pairs of nodes
in the network, we could discover network capacity for transit using the
standard graph theoretic flow network algorithm, for example, the Ford-
Fulkerson algorithm (Kleinberg and Tardos, 2005). Transit/flow rates
incur a cost corresponding to the amount of flow that needs to be paid
by the pair of a sender and a receiver. It may be beneficial to share the
transmission cost with neighbors, who form game theoretic coalitions
that will be discussed in Chapter 3.
1.7 CONCLUSIONS
Networks are abundantly around us. They are man-made or naturally
occur. They are implicit, hidden, explicit, or articulated. They might be
tangible and objectively quantified, or they might be subjective and dif-
ficult to quantify. They all tend to change in time, which is the subject of
our future chapters on network dynamics.
REFERENCES
Albert, R., Barabási, A.-L., 2002. Statistical mechanics of complex networks. Rev. Mod. Phys. 74,
47–97.
Barabási, A.L., Albert, R., 1999. Emergence of scaling in random networks. Science 286, 509–512.
Carmona, G., 2012. Existence and Stability of Nash Equilibrium. World Scientific Publishing
Company.
Epstein, D., Wang, J., 2002. A steady state model for graph power laws. In: Proceedings of 2nd
International Workshop on Web Dynamics. World Scientific Publishing Company.
Erdős, P., Rényi, A., 1959. On random graphs. Publicationes Mathematicae 6, 290–297.
Gabaix, X., 2009. Power laws in economics and finances. Annu. Rev. Econ. 1, 255–294.
Jackson, M., 2003. A survey of models of network formation: stability and efficiency. In: Demange,
G., Wooders, M. (Eds.), Group Formation in Economics: Networks, Clubs, and Coalitions.
Cambridge University Press.
Jornet, J.M., Pierobon, M., 2011. Nanonetworks: a new frontier in communications. In: Communications
of the ACM. Vol. 54, No. 11. ACM, pp. 84–89.
Khare, P., 2012. Social Media Marketing eLearning Kit For Dummies. Wiley.
Kleinberg, J., 2000. The small-world phenomenon: an algorithmic perspective. In: Proceedings of
32nd ACM Symposium on Theory of Computing. ACM, pp. 163–170.
Kleinberg, J., 2001. Small-world phenomena and the dynamics of information. In: Proceedings of
the Advances in Neural Information Processing Systems (NIPS), Vol. 14. NIPS.
Kleinberg, J., Tardos, E., 2005. Algorithm Design. Addison-Wesley.
Li, W., 1992. Random texts exhibit Zipf’s-law-like word frequency distribution. IEEE Trans. Inf.
Theory 38 (6), 1842–1845.
McPherson, M., Lovin, L.S., Cook, J., 2001. Birds of a feather: homophily in social networks.
Annu. Rev. Sociol. 27, 415–444.
Milgram, S., 1967. The small world problem. Psychol. Today 1 (1), 61–67.
Moffett, M., 2010. Adventures Among Ants. University of California Press.
Nielsen, M., 2012. Reinventing Discovery: A New Era of Networked Science. Princeton University
Press.
Penrose, M., 2003. Random Geometric Graphs. Oxford University Press.
Reingold, H., 2000. The Virtual Community. MIT Press.
Seung, S., 2012. Connectome: How Brain’s Wiring Makes Us Who We Are. Mariner Books.
Watts, D., Strogatz, S., 1998. Collective dynamic of small-world networks. Nature 393 (6684), 440–442.
14 Computational Network Science: An Algorithmic Approach
EXERCISES
1. Using examples, describe how animal swarms are networked.
2. What are the salient characteristics of biological networks (e.g.,
brain cells and protein chains) that differentiate them from other
types of networks?
3. What will be the role of network organizations in the year 2025?
Give examples.
4. How can social media be used to track cultural changes in a society?
CHAPTER 2
Network Analysis
There has been a long tradition of measuring qualities for network
locations from both egocentric and global perspectives. This is largely
addressed with quantification attempts in mathematical sociology under
the theme of social network analysis (SNA) (Wasserman and Faust, 1994;
Knoke and Yang, 2007; Golbeck, 2013; Borgatti et al., 2013). There
are also several popular software toolkits that perform analysis and
visualization of social networks (i.e., sociograms) including UCINET
and NodeXL. Tom Snijders’ SIENA is a program for the statistical
analysis of network data. The NSF-sponsored visualization project is
Traces (Suthers, 2011), which traces out the movements, confluences,
and transformations of people and ideas in online social networks.
The aim of this chapter is to review a selective subset of SNA
measures that complement algorithmic descriptions explained in the
remainder of this book. For a glossary of SNA terms, readers are
recommended to consult Golbeck (2013).
We will start with egocentric (i.e., node view) measures. A degree-1
network of a node is the node and its immediate neighbor nodes. A
degree-1.5 network of a node is the node’s degree-1 network and its links
among immediate neighbors (Golbeck, 2013). A degree-2 network of
a node is the node’s degree-1 network and all its immediate neighbors’
connections (Golbeck, 2013). A degree-n network of a node is the
degree-1 network of the node plus all the nodes and the corresponding
links that are no more than n links away from the starting node.
A path is a chain (i.e., succession) of nodes connected by links be-
tween pairs of nodes. Two nodes are connected if and only if (i.e., iff)
there is a path between them. A connected component is a set of nodes
with connected paths among all pairs of nodes in the set. A bridge is a
link that connects two isolated connected components. A hub is a node
with many connections. Reachability is whether two nodes are connected
or not by way of either a direct or an indirect path of any length.
16 Computational Network Science: An Algorithmic Approach
RD kj
Radiality ( k ) = ∑ (2.2)
j ≠k n −1
6 × number of triangles
B= (2.6)
number of length between two paths
REFERENCES
Bonacich, P., Lu, P., 2012. Introduction to Mathematical Sociology. Princeton University Press.
Borgatti, S., Everett, M., Johnson, J., 2013. Analyzing Social Networks. SAGE Publications Ltd.
Burt, R., 1995. Structural Holes: The Social Structure of Competition. Harvard University Press.
Chiang, M., 2012. Networked Life: 20 Questions and Answers. Cambridge University Press.
Freeman, L., 1978. Centrality in social networks: conceptual clarification. Soc. Netw. 1, 215–239.
Golbeck, J., 2013. Analyzing the Social Web. Morgan Kaufmann.
Knoke, D., Yang, S., 2007. Social Network Analysis. Sage Publications.
Macindoe, O., Richards, W., 2011. Comparing networks using their fine structure. Int. J. Soc.
Comput. Cyber-Phys. Syst. 1 (1), 79–97, Inderscience Publishers.
Suthers, D., 2011. Interaction, mediation, and ties: an analytic hierarchy for socio-technical
systems. In: Proceedings of the Hawaii International Conference on the System Sciences
(HICSS-44). January 4–7, 2011, Kauai, Hawai‘i.
Valente, T., Foreman, R., 1988. Integration and radiality: measuring the extent of an individual’s
connectedness and reachability in a network. Soc. Netw. 20 (1), 89–105.
Wasserman, S., Faust, K., 1994. Social Network Analysis: Methods and Applications. Cambridge
University Press.
Watts, D., Strogatz, S., 1998. Collective dynamics of ‘small-world’ networks. Nature 393 (6684),
440–442.
EXERCISES
1. How can we track search queries (i.e., say in Google or YouTube)
from social network profiles of corresponding users?
2. Describe a network that appears to be but is not a scale-free
network.
3. How can beta centrality measure be used to find weakly connected
nodes in a network?
4. Modify density metric for weighted networks, that is, networks with
links that have weights as strengths of connections.
CHAPTER
Network Games
3
Decision making requires reasoning. Whereas decision theory is about the
process of an individual’s reasoning processes when pertinent decision
attributes can be independently ascertained, game theory (GT) is about
the process of reasoning when pertinent decision attributes include
decisions of other individuals (Fudenberg, 1991). The latter is the scenario
in networks where all the decisions are interdependent. GT has been
a branch of mathematics (Barron, 2008) and has long been used to
explain economic decision making in the theories of microeconomics
(Mas-Colel et al., 1995). We will briefly introduce GT in Section 3.1 before
the discussion of network-relevant applications in Sections 3.2–3.6.
µi = si × p(S1 + S2 ) − ci × Si (3.1)
For Cournot games, it is typical to plot the best responses of two
players (i.e., strategies with optimal payoffs for players) with S1 versus
S2. In such plots, the point of intersection of the two best response lines
denotes the equilibrium point (i.e., stability point) often denoted as S*
(i.e., the amount of goods either firm should produce) where neither
player will have an incentive to unilaterally abandon the strategy pre-
scribed by the equilibrium.
A famous two-player competitive game is Prisoner’s Dilemma (PD)
with prototypical payoffs shown in Figure 3.2 with two strategies of
cooperation (C) and defection (D). In PD games, D is the dominant
strategy (i.e., the strategy that yields higher payoff for the player)
regardless of players’ choices. Often, there are strategies that might be
dominated (e.g., suicide would always produce a loss–loss strategy com-
bination in PD) and game analysis often suggests elimination of such
strategies (Myerson, 1997). Nash equilibrium (NE) for a game 〈I, S, U〉 is
a strategy profile S* ∈ S (i.e., an ideal strategy combination for players)
such that for all i ∈ I and for all Si ∈ S, Equation 3.2 holds. NE is a
form of equilibrium (i.e., stability). In many competitive games, there
are multiple equilibria, among which we must select the most desirable
one based on contextual biases (Nisan et al., 2007). A measure of equi-
librium efficiency is the Price of Anarchy that is the ratio between the
worst and the best equilibria (Roughgarden, 2005).
DEUXIÈME PARTIE