=Paper=
{{Paper
|id=Vol-1830/Paper53
|storemode=property
|title=A Survey of Influential Nodes Detection Methods in Mobile Phone Network
|pdfUrl=https://ceur-ws.org/Vol-1830/Paper53.pdf
|volume=Vol-1830
|authors=Elizabeth N. Onwuka,Bala A. Salihu,Sheriff Murtala
}}
==A Survey of Influential Nodes Detection Methods in Mobile Phone Network==
https://ceur-ws.org/Vol-1830/Paper53.pdf
International Conference on Information and Communication Technology and Its Applications
(ICTA 2016)
Federal University of Technology, Minna, Nigeria
November 28 – 30, 2016
A Survey of Influential Nodes Detection Methods in Mobile Phone Network
Elizabeth N. Onwuka, Bala A. Salihu, and Sheriff Murtala
Department of Telecommunications Engineering, Federal University of Technology, Minna, Nigeria
Abstract—The number of mobile phone users is increasing CDR contains metadata that describes a specific instance
tremendously. The social interaction between these mobile of a telecommunication transaction (calls, messages and
phone users can be represented using social network graphs. Internet services) but does not include the content of that
This type of study has very important applications in various transaction, for example, CDR for a particular call contains
areas especially in the detection of criminal groups who also both the caller and receiver‘s number, the time stamp (date
use these devices to interact and plan their activities. and time), the duration of the call, the base station ID of the
Moreover, the study of identifying influential nodes in social caller‘s location and other related information. CDR may
network of any kind is currently receiving attention in the capture thousands or millions of users within a specific time
research arena. This is because identification of influential
and place and it can be used to create a network of mobile
nodes of any network is significant to understanding the
network. This becomes very important if the network in
phone subscribers. CDR is a huge repository of human
question is a criminal network, considering the insecurities of behavioural data and it belongs to the group of data being
the current time. In this paper, a survey of influential nodes currently described as Big Data. The information from CDR
detection methods is carried out, we first define the problems reveals the inter-relationship network between mobile phone
associated with influential nodes detection and then examine subscribers at various spheres, generally called social
various methods of identifying influential nodes. We also networks. A mobile phone network is a social structure that
consider techniques employed in analysing users in the mobile represents the interconnection of mobile phone subscribers
phone network. based on call detail record (CDR). An example of a social
network of mobile phone users is shown in Fig. 1. The idea
Keywords—social network; mobile phone network; of forming a social interaction between mobile phone users
influential nodes detection; centrality measures; support researchers in the different area of studies like
personal mobility prediction, fraud detection in
telecommunication [2], urban planning and development,
geographical partitioning [3] and intelligence gathering for
national security [4].
I. INTRODUCTION Human beings normally form groups or clusters based on
Since the invention of mobile communication and other certain commonalities. These groups (called communities)
services attached to it, many people find it better and cheaper also reflected on the communication data. Networks are
to communicate using the medium than wired made up of communities and in each community, there are
communication thereby attracting more subscribers to use nodes with varying degrees of influence, these nodes are
mobile communication network. A survey carried out by called influential nodes. A good area of application of mobile
international telecommunication union (ITU) shows that the phone network is in the detection of influential mobile
population of mobile phone subscribers increased from 738 subscribers. For instance, a network of mobile phone
million in the year 2000 to 7 billion in 2015 and within this subscribers which is created by collecting call record
same time the proportion of population covered by a 2G information from a reasonable number of actors (that act as
mobile cellular network increased from 58% to 95% with seeds) will consist of different communities and some users
more remote areas captured [1]. In developing countries, at within these communities will influence other users either
least one member of every household communicates using a positively or negatively. The major problem in this area is
mobile phone. Each subscriber enjoys making calls and how to accurately determine the genuine influential
receiving calls from other users and enjoys the same for short individuals in a social network. In this paper, we present an
messages and Internet services. Telecommunication overview of various ways of finding influential nodes in a
networks have really made the world a global village in the social network.
sense that peoples‘ social reach has expanded even across The remainder of this paper is organised as follows:
borders. The log of activities of each user is stored on the Section II provides a brief background and review of related
user‘s phone and also recorded with the Mobile Network works. Section III describes different methods of identifying
Operators (MNOs). The information collected by the MNOs influential nodes in a mobile phone network. Finally, we
is referred to as Call Detail Record (CDR). conclude this paper in Section IV.
213
International Conference on Information and Communication Technology and Its Applications (ICTA 2016)
in the social network. This is added to the fact that
researchers and investigators have taken full advantage of
social network analysis to unravel the operation of terrorists
and criminals [4]. Crime investigation application becomes
more necessary now that communication networks have
changed the way people live and transact business. It is
intuitively believed that criminals rely on this network for
planning criminal activities of all sorts.
In our study, we focus on identifying important and
interesting nodes in a mobile phone network and we discuss
some of the previous studies that have been done in this
research area by first looking at the major problems and
concepts employed in the detection of important nodes and
different approaches that had been applied so far.
Vertex list, V Edge List, E Weight, W
A C B 10
B F G 15
C D F 3
D A E 12
E E D 7
F A F 13
Figure 1. A snapshot of a mobile phone network sample with circles
indicating mobile phone users and edge weight colour coded from yellow G C E 4
(weak link) to red (strong link) [5] E F 9
A B 9
A G 12
II. RELATED WORK B E 14
C A 8
2.1 Background D B 11
There is a rapidly growing literature on influential nodes Graph, G = (V,E,W)
discovery in social networks, which indicates that a lot of
study had been carried out in this field [6], [7], [8], [9], [10],
[11], [12], [13], [14], [15], [16], [17]. However, due to the
challenges of getting mobile phone data, little studies have Figure 2. Modelling graph from a set of vertex, edge and weight.
been carried out on discovering communities and important
mobile subscribers in the mobile phone network. A mobile 2.2 Influential Nodes Detection Problem
phone network is treated like any other social network that
has a tree network structure. Social network is usually Influential nodes are set of nodes whose roles are very
modelled as a graph, G=(V,E), where V is a set containing important in the spread of influence across the network.
all nodes (actors) in the network and E is also a set These nodes have the tendency to influence other nodes
containing all edges (links) between two elements (pairs) of either constructively or destructively. Influential nodes and
set V. If the direction of the edges is considered the graph is ―key nodes‖ seem to be the same. According to Borgatti [6],
said to be directed or undirected if otherwise. Also, when the influential nodes‘ problem can either be a key player
corresponding weight, W of the edges is considered, the problem positive (KPP-Pos.) or key player problem negative
graph, G=(V,E,W) is said to be weighted or binary (KPP-Neg.). KPP-Pos. is defined with respect to the way key
(unweighted) if otherwise. A simple description of how an nodes are connected and integrated into the network, while
undirected and weighted graph is modelled from set V, E and KPP-Neg. is defined in relation to the network reliance on its
W is shown in Fig. 2. key nodes to sustain its connectedness.
Exploring social network data requires basic concepts of Recently, [7][16] presented an overview of existing
graph representation, analysis and visualisation [18]. These techniques of finding important and influential nodes in
concepts include centrality measures, shortest path problems, social networks. In this paper, we extend the study of Probst
clustering techniques and network density. This is necessary by reviewing more novel approaches in finding prominent
when interpreting the result in order to have a good nodes in the social network with emphasis on mobile phone
understanding of the social interactions between nodes in a network. For clarity, we classify some of the previous work
network. Due to the rich resources in social network on influential nodes detection into two categories: centrality
analysis, it serves as a tool for analysing and visualising big and non-centrality measures.
data [19]. Some major areas of study in the social network 1) Centrality measures
analysis are community structure, detection of cliques and In graph theory and network analysis, the most important
discovery of key nodes and neighbours. Recently, more tool is centrality measure. Centrality measures are
attention has been given to the detection of influential nodes considered as structural measures of influence that indicate
214
International Conference on Information and Communication Technology and Its Applications (ICTA 2016)
a user‘s position in a social network. Degree centrality, bridge along the shortest path between two other nodes. The
betweenness centrality, closeness centrality, and eigenvector betweenness centrality of a user, v is given by
centrality are the four widely used centrality measures in st (v)
determining the relative importance of a user within a BC (v)
s t vV st
(4)
network. Although these measures have limitations, they
have been proven to be the basis of other methods of
identifying key nodes based on specific purposes within a where st is the number of shortest paths in the graph, G
social network [20]. between nodes ―s‖ and ―t‖; st (v) is the number of
shortest paths in the graph, G between nodes ―s‖ and ―t‖ that
a) Degree Centrality: is defined as the number of
pass through user ―v‖. Nodes with high betweenness are
edges incident upon a user. In other words, this measure
responsible for controlling the spread of information across
indicates how many nodes can be directly reached by a
the graph. However, they might not be responsible for
particular node. The degree centrality of a user, v is given
causing maximum disconnection (fragment) within the
by:
network [6]. Brandes also presented an algorithm for
computing the betweenness index of a large number of
DC (v) deg(v) (1) nodes [25].
d) Eigenvector Centrality: Eigenvector centrality (also
deg(v, G) |{u V : (u, v) E}| (2) called eigencentrality) is a measure of how well a particular
user is connected to other influential nodes. This is one of
Nodes with high degree centrality score might be the oldest centrality measures developed to assist the social
considered influential. The flaw of this centrality measure is analyst in recognising the behaviour of people [26]. To
that it relies on direct connections between nodes. Using this determine eigenvector centrality, it is imperative to first find
individual centrality alone to determine the key nodes will the adjacency matrix, A of the graph, G. Given
result in the selection of nodes that only have a high number A a(v, u) for a binary network.
of direct connections.
b) Closeness Centrality: According to Bavelas, Where:
closeness centrality of a user as the reciprocal of the sum of
its distances from all other nodes [21]. This measure is
0,
1, if the link between node v and u exist ,
effective in describing the hierarchy among members of a a v, u
group and can also be used to indicate how fast a user can otherwise.
reach every other user in the network. The closeness
centrality of a user, v is given by: The eigenvector centrality of a node, v is mathematically
defined as:
1 1
CC (v) d (v, u ) (3)
xv x
uM ( v ) u
(5)
u
The boundary of the summation, is all members of the
where (u, v) E and d (v, u ) is the sum of the length set of neighbours of v, M(v). In matrix representation,
eigenvector centrality is expressed mathematically as:
of all shortest paths from
u all other nodes from node ―v‖.
Okamoto et al. [22] introduced an efficient algorithm for
discovering the top k-highest closeness centrality nodes 1
xv Axu (6)
called TOPRANK. The algorithm is made up of the
approximate algorithm and exact algorithm. The
approximate algorithm is applied to identify the top nodes Where is the eigenvalue (constant) and xv is the
with high centrality scores while the exact algorithm is used corresponding eigenvector of the adjacency matrix, A.
to rank the detected nodes. [23] presented a closeness Eigenvector centrality is much related to Katz centrality
centrality algorithm that efficiently determines the closeness [27], a universality of degree centrality. Katz centrality
scores of each user any time the social network structure is measures the relative influence of nodes within a social
modified. The changes involve the insertion of a new edge network by determining the number of the immediate
or the removal of an existing edge. neighbours (first-degree nodes) and also all further nodes in
c) Betweenness Centrality: Betweenness-based the network that connect to the node under consideration
centrality measures were first introduced by Freeman in through these close neighbours.
[24]. The author discovered that it is important to generalise e) Other Centrality-Based Approaches: The number of
the concept of point centrality and structural properties of centrality measures extends beyond the four metrics
the social network from past study[21]. Betweenness discussed earlier. It is quite interesting that most of the new
centrality expresses the number of times a user acts as a measures were related one way or the other to the four most
215
International Conference on Information and Communication Technology and Its Applications (ICTA 2016)
popular centrality measures with a little modification. closeness centrality, betweenness centrality and eigenvector
Stephenson and Zelen suggested a new centrality measure centrality to determine the influence factor of actors in an
called Information centrality [28]. This centrality has online social network obtained from Facebook [9]. The study
statistical theory background that considers all the path shows that these three centrality measures are important in
signals in a social network. Although this centrality is measuring the influence of each user and as well as the
applied to undirected networks only, it supersedes other influence of the entire social network. Srinivas and
traditional centrality measures in identifying the most Velusamy [10] presented an algorithm that combines degree
central nodes. In [29], the authors proposed an optimal inter- centrality and clustering coefficient to discover influential
centrality measure, which takes into account both the user‘s nodes in three different datasets collected from Facebook.
The clustering coefficient feature is used to enhance the
centrality and its impact on the centrality of the other nodes.
traditional degree centrality. According to their study, the
Using a criminal network, the authors‘ findings indicated nodes with the least degree indicated that they are more
that the key criminal is criminal with the highest optimal connected and thus, the most influential. The algorithm is
inter-centrality in the network and the removal of this found to be effective in discovering important nodes. In [38],
criminal would greatly reduce the crime rate. In [30] and a criminal network is constructed and analysed using
[31], the authors extended the work of Ballester et al., by PageRank and other centrality measures to identify key
presenting an inter-centrality with key group dimension. criminals.
The modified inter-centrality explores the key group whose
2) Non-centrality measures
nodes are different from the nodes with highest individual In this subsection, we consider previous studies that
inter-centralities. employed other techniques different from centrality
Ilyas and Radha introduced a new centrality called approach in detecting influential nodes.
principal component centrality (PCC), a variant of a) Information theory: Shetty and Adibi [39] presented
eigenvector centrality [32]. PCC is based on principal an information theory approach called graph entropy to
component analysis (PCA) and karhunen loeve transform
discover the dominant nodes within a social network. The
(KLT) which handles graph adjacency matrix as a
covariance matrix. Contrary to eigenvector centrality, PCC graph entropy of the entire network is determined every
provides more features for centrality computation. Moreover, time a user is removed from the network. The removal of
an investigation was carried out to detect influential nodes in nodes that cause great disorder in the graph entropy are seen
two separate datasets using eigenvector centrality and as influential nodes. Furthermore, [40] described an entropy
principal component centrality [33]. Their results showed measure based on Shannon measure of uncertainty. This
that eigenvector centrality considered the most influential entropy is applied to networks whose traffic moves along
node within the largest community in a network and paths by transference. One great advantage of this measure
consequently ranks the neighbours of the influential node is nodes with betweenness score of zero can actually have
and ignores other nodes in the remaining small communities reasonable entropy values. In another developmental study,
that have low eigenvector scores. In the case of PCC, it Ortiz-Arroyo et al. [41] proposed two entropy-based
considered both the nodes in the largest community and measures called connectivity entropy and centrality entropy.
other nodes with zero eigenvalues in small communities. The authors further demonstrated how the entropy-based
Despite the introduction of these new centrality measure can effectively solve the two key player problems
measures, the fact still remains that an individual centrality [6]. Although their result is similar to that of combinatorial
measure might not be the most appropriate for identifying optimisation algorithms the major setback for entropy
influential nodes in a social network. A centrality measure is approach is it cannot work on large graphs. The approach is
applied depending on a specific role of nodes in the network. computationally difficult to implement because it requires a
For instance, nodes that are influential gossipers or most
path finding algorithm to simplify the operation of the
spreaders of virus function as information regulators in the
entropy centrality.
network. Another different purpose is identifying nodes that
can maximally disrupt the social network. The irregularities b) Activity Based: This measure focuses on the
in some of these individual centrality measures have open up activities between nodes in a social network. Goldenberg et
fascinating research fields on group and improved centrality al.[42] claimed that some individuals are more important
measures that can be universal in identifying the most than others based on their activity. The authors proposed an
influential nodes [34][35]. Some studies also considered activity based measure for identifying influential nodes by
combining two or more centralities measures in getting a selecting hub with high in-degree and out-degree in a
general set of influential nodes. Sathik and Rasheed directed network. Heidemann et al. revealed that not all
proposed an algorithm to identify sets of key players based connection links are active in a social network and the
on centrality measures [36]. The authors addressed the key active links are insignificant [43]. They proposed an
player problems [6], using closeness centrality, degree undirected activity based measure by modifying the
centrality and betweenness centrality. Zaman [37] in his PageRank algorithm for discovering important nodes in the
(unpublished) PhD thesis recommended a new centrality
network. The modified PageRank also considers the
measure known as rumor centrality that determines the
source of rumor (gossip) and how influence (gossip) spread weighted edges between the nodes.
across a microblogging website (twitter). c) User Preferences and Attributes: Zhang et al.[11]
Lately, in order to adequately discover real influential developed an algorithm while trying to solve the influence
nodes. Ahsan et al. described a scheme that combines maximisation problem called greedy algorithm based on
216
International Conference on Information and Communication Technology and Its Applications (ICTA 2016)
user preferences (GAUP). The algorithm is applied along machine learning technique that aims to find out how one
with an extended independent cascade (EIC) model and item affects another by analysing how frequently certain
takes the user preferences into the influence diffusion items appear together in a specific dataset [15]. ‗Association
process, thus making it the first algorithm that mines the rule learning is carried out by applying two norms, namely,
top-k influential nodes based on user preferences while support and confidence. Support specifies the proportion of
trying to solve influence maximisation problem. This such items, while confidence specifies how many times
algorithm surpasses the existing greedy algorithm that uses those rules in the whole dataset are accurate. The influential
independent cascade (IC) model. Canali and Lancellotti [44] nodes are listed as nodes with a confidence level of 95% and
presented a numerical method of detecting key nodes in a above. The technique is easy to implement and proven to be
social network by considering the nodes‘ attributes similar to PageRank and degree centrality.
(personal information, nodes preferences, activities,
uploaded contents, content accesses). The desired attributes
of the nodes were collected and summed up using principal III. IDENTIFYING INFLUENTIAL NODES IN A MOBILE PHONE
component analysis (PCA). The dimension of the PCA is NETWORK
matched with the nodes attributes and the dimension that In this section, research methods that are applied in
greatly reflect the nodes‘ attributes is considered in defining detecting influential nodes in mobile phone networks were
a metric that determines the influential nodes in the social discussed. Over time, researchers attempted to study and
network. For each metric selected, the authors ranked the analyse Call Detail Record (CDR) [5][45]. However, Mobile
nodes and selected the top influential nodes accordingly. Network Operator(s) (MNOs) are strongly reluctant to
release mobile phone data to the public due to privacy issue.
d) Profile based characteristics: Nodes interactions are In cases where CDR is released to third parties, MNOs might
characterised as either popular, active or both. Eirinaki et al. conceal the identity of their users or non-disclosure
introduced a measure named ProfRank [12], that uses the agreement (NDA's) contract is involved in protecting
popularity and activity characteristics to identify the most customers' privacy. But if the agreement fails, another way to
influential nodes within a social network. The authors also collect CDR from mobile phone users is to develop a
showed that the measure performed better than betweenness programme that extract users‘ log of activities. The
centrality and Pagerank. programme is usually installed on user‘s mobile phone and
each information retrieved from the user is stored in a
e) Influence Graph: Agarwal et al.[13] argued that dedicated database [46]. The process is expensive and takes
influential bloggers are not necessarily active bloggers longer time but the result worth it.
(nodes) on a blogging website. The authors defined Kiss and Bichler [47] compared the performances of
statistical properties (recognition, activity generation, seven existing centrality measures including SenderRank, a
novelty and eloquence) that are related to influence between new technique which was developed by the authors, to
a blog post and proposed an influence graph model that identify influential users in a social network constructed
measures the influence of each blog post across the from a dataset collected from a telecom company.
community blog site. The influence is dependent on the SenderRank and Out-degree centrality performed well in
determining the most central nodes in a network of calls
influence flow and additive weight function that regulates a
from a telecommunication company. In [48], degree
number of comments. The influence score is assigned to
centrality and betweenness centrality were combined with
each blog‘s post. The maximum influence score is selected various seeding strategies to discover prominent nodes in an
as the reference and its influence score as the blogger index. anonymized mobile phone network and two other social
Using this index, the bloggers are ranked accordingly to the networks.
index and the most influential bloggers are the top ranked. Catanese et al. [49] proposed a tool called LogAnalysis
f) ShaPley value-based: Narayanam and Narahari, for scientific analysis of real phone call networks. This social
while trying to solve the influence maximisation problem network analysis tool provides both statistical and visual
proposed an algorithm called ShaPley value-based representation of real mobile phone network. Different
centrality measures are featured in the statistical operation of
Influential Nodes (SPINs)[14]. Eventually, SPINs solved
this tool and they are used in ranking users according to how
the top-k nodes problem and -coverage problem. The top- important they are in the phone call networks. The
k problem involves the discovery of a set of influential application of this tools is not only restricted to phone call
nodes for maximising the spread of information. The - networks but can also be applied in investigating criminal
coverage problem focuses on the identifying the set of networks [50].
influential nodes having least cardinality with which it is Han et al. [17] presented a program called iWander that
possible to influence a fixed percentage, of the nodes in runs on the mobile device of users (nodes). Random walk
the social network through the process of diffusion. The messages are sent to the users at fixed length of time.
authors showed that SPIN is computationally efficient when iWander determines the most influential users by computing
the centrality of each node based on the total random walk
compared to other existing greedy algorithms.
messages received by each node. The authors carried out a
g) Association Rule Learning: Erlandsson et al. theoretical analysis of the program and showed that
discovered influential nodes in a social network by applying influential nodes identified by iWander can regulate the
asocial rule learning. ―Association rule learning is a spread of communicable diseases and can further be used to
217
International Conference on Information and Communication Technology and Its Applications (ICTA 2016)
avoid a total epidemic. We summarised these methods in [10] A. Srinivas and R. L. Velusamy, ‗Identification of influential users
Table 1. from social networks based on Enhanced Degree Centrality Measure‘,
in Advance Computing Conference (IACC), 2015 IEEE International,
2015, pp. 1179–1184.
TABLE I. INFLUENTIAL NODES DETECTION TECHNIQUES IN MOBILE
PHONE NETWORK [11] Y. Zhang, Z. Wang, and C. Xia, ‗Identifying Key Users for Targeted
Marketing by Mining Online Social Network‘, 2010, pp. 644–649
Mobile [12] M. Eirinaki, S. P. S. Monga, and S. Sundaram, ‗Identification of
Authors Technique Components phone influential social networkers‘, Int. J. Web Based Communities, vol. 8,
Dataset no. No. 2, pp. 136–158, 2012.
Centrality In-degree, out-degree, [13] N. Agarwal, H. Liu, L. Tang, and P. S. Yu, ‗Identifying the influential
Kiss and
measures and betweenness, closeness, Yes bloggers in a community‘, in Proceedings of the 2008 international
Bichler[25]
non-centrality Pagerank, SenderRank conference on web search and data mining, 2008, pp. 207–218.
Centrality
Hinz et al., [27] measures and
Centrality measures and
Yes [14] R. Narayanam and Y. Narahari, ‗A Shapley Value-Based Approach to
non-centrality
seeding techniques. Discover Influential Users in Social Networks‘, IEEE Transactions
on Automation Science and Engineering, vol. 8, no. 1, pp. 130–147,
Catanese et al., LogAnalysis
Centrality measures. Yes Jan. 2011.
[46]
Centrality measures [15] F. Erlandsson, P. Bródka, A. Borg, and H. Johnson, ‗Finding
Han et al., [48] iWander based on random walk Yes Influential Users in Social Media Using Association Rule Learning‘,
sampling. Entropy, vol. 18, no. 5, p. 164, Apr. 2016.
[16] S. Singh, N. Mishra, and S. Sharma, ‗Survey of various techniques
for determining influential users in social networks‘, in Emerging
Trends in Computing, Communication and Nanotechnology (ICE-
IV. CONCLUSION CCN), 2013 International Conference on, 2013, pp. 398–403.
We presented a background study on the detection of [17] B. Han, J. Li, and A. Srinivasan, ‗Your friends have more friends than
you do: Identifying influential mobile users through random-walk
influential nodes and considered the key player problems as sampling‘, IEEE/ACM Transactions on Networking, vol. 22, no. 5,
our focus in this study. We also discussed some of the pp. 1389–1400, 2014.
previous studies related to discovering the most important [18] A. Abraham, Ed., Computational Social Networks. London: Springer
and interesting nodes in a social network. We observed that London, 2012.
centrality measures are more general and effective in the [19] M. Lieberman, ‗Visualising big data: Social network analysis‘, in
detection of influential nodes in social networks and mobile Digital research conference, 2014.
phone network. Though, results of using non-centrality [20] D. M. B. Friedl and J. Heidemann, ‗A critical review of centrality
measures are comparable to centrality measures approach. measures in social networks.‘, Business & Information Systems
Thus, the idea of combining two or more centrality measures Engineering, vol. 2, no. 6, pp. 371–385, 2010.
would improve detection and solve the influential nodes‘ [21] A. Bavelas, ‗Communication patterns in task-oriented groups., 22(6)‘.
J. Acoust. Soc. Am, 1950.
detection problem.
[22] K. Okamoto, W. Chen, and X.-Y. Li, ‗Ranking of closeness centrality
It would be quite interesting if this study can cover other for large-scale social networks‘, in International Workshop on
areas of communication network of wireless sensors, internet Frontiers in Algorithmics, 2008, pp. 186–195.
hubs and access points. [23] Ahmet Erdem Sarýyu¨ce, Kamer Kaya, Erik Saule, and U¨ mit V. C¸
atalyu¨rek, ‗Incremental closeness centrality in distributed memory.‘,
REFERENCES Parallel Computing, vol. 47, pp. 3–18., 2015.
[24] L. C. Freeman, ‗A Set of Measures of Centrality Based on
[1] ‗The world in 2014: ICT Facts and Figures. International
Betweenness‘, Sociometry, vol. 40, no. 1, p. 35, Mar. 1978.
Telecommunication Union.‘
[25] U. Brandes, ‗A faster algorithm for betweenness centrality*‘, Journal
[2] C. A. R. Pinheiro, ‗Community Detection to Identify Fraud Events in
of mathematical sociology, vol. 25, no. 2, pp. 163–177, 2001
Telecommunications Networks‘, 2012.
[3] V. D. Blondel, A. Decuyper, and G. Krings, ‗A survey of results on [26] J. R. Seeley, ‗The net of reciprocal influence: A problem in treating
mobile phone datasets analysis‘, EPJ Data Science, vol. 4, no. 1, p. 1, sociometric data.‘, Canadian Journal of Psychology, vol. 3, pp. 234–
2015. 240, 1949.
[27] L. Katz, ‗A new status index derived from sociometric analysis‘,
[4] J. D. Farley, ‗Breaking Al Qaeda Cells: A Mathematical Analysis of
Psychometrika, vol. 18, no. 1, pp. 39–43, 1953.
Counterterrorism Operations (A Guide for Risk Assessment and
Decision Making)‘, Studies in Conflict & Terrorism, vol. 26, no. 6, [28] K. Stephenson and M. Zelen, ‗Rethinking centrality: Methods and
pp. 399–411, Nov. 2003. examples‘, Social Networks, vol. 11, no. 1, pp. 1–37, 1989.
[5] J.-P. Onnela, J. Saramäki, J. Hyvönen, G. Szabó, M. A. De Menezes, [29] C. Ballester, A. Calv´o-Armengol, and Y. Zenou, ‗―Who‖s who in
K. Kaski, A.-L. Barabási, and J. Kertész, ‗Analysis of a large-scale crime networks. Wanted: the key player‘.‘, Discussion Paper 4421.,
weighted network of one-to-one human communication‘, New vol. 4421, no. CERP, 2004.
Journal of Physics, vol. 9, no. 6, p. 179, 2007 [30] S. Sarangi and E. Unlu, ‗Key players and key groups in teams: A
[6] S. P. Borgatti, ‗Identifying sets of key players in a social network‘, network approach using soccer data.‘ 2010.
Computational and Mathematical Organization Theory, vol. 12, no. [31] U. Temurshoev, ‗Who‘s Who in Networks-Wanted: The Key Group‘,
1, pp. 21–34, Apr. 2006. Available at SSRN 1285752, 2008.
[7] F. Probst, ‗Customer Relationship Management in a Digitally [32] M. U. Ilyas and H. Radha, ‗A KLT-inspired user centrality for
Connected World‘, 2013. identifying influential neighborhoods in graphs‘, in Information
[8] X. Zhang, J. Zhu, Q. Wang, and H. Zhao, ‗Identifying influential Sciences and Systems (CISS), 2010 44th Annual Conference on, 2010,
users in complex networks with community structure. Knowledge- pp. 1–7.
Based Systems‘, Knowledge-Based Systems, vol. 42, pp. 74–84, 2013 [33] M. U. Ilyas and H. Radha, ‗Identifying influential users in online
[9] M. Ahsan, T. Singh, and M. Kumari, ‗Influential user detection in social networks using principal component centrality‘, in
social network during community detection‘, in Cognitive Computing Communications (ICC), 2011 IEEE International Conference on,
and Information Processing (CCIP), 2015 International Conference 2011, pp. 1–5
on, 2015, pp. 1–6.
218
International Conference on Information and Communication Technology and Its Applications (ICTA 2016)
[34] M. G. Everett and S. P. Borgatti, ‗The centrality of groups and [42] J. Goldenberg, S. Han, D. R. Lehmann, and J. W. Hong, ‗The role of
classes‘, The Journal of mathematical sociology, vol. 23, no. 3, pp. hubs in the adoption process‘, Journal of Marketing, vol. 73, no. 2,
181–201, 1999. pp. 1–13, 2009
[35] A. Sheikhahmadi, M. A. Nematbakhsh, and A. Shokrollahi, [43] J. Heidemann, M. Klier, and F. Probst, ‗Identifying key users in
‗Improving detection of influential nodes in complex networks‘, online social networks: A PageRank based approach‘, 2010.
Physica A: Statistical Mechanics and its Applications, vol. 436, pp. [44] C. Canali and R. Lancellotti, ‗A quantitative methodology based on
833–845, Oct. 2015. component analysis to identify key users in social networks‘,
[36] M. M. Sathik and A. A. Rasheed, ‗A centrality approach to identify International Journal of Social Network Mining, vol. 1, no. 1, pp. 27–
sets of key players in an online weblog‘, International Journal of 50, 2012.
Recent Trends in Engineering, vol. 2, 2009. [45] F. C. Perkins III, System and method for processing call detail
[37] T. R. Zaman, ‗Information extraction with network centralities: records. Google Patents, 2002
finding rumor sources, measuring influence, and learning community [46] A. McDiarmid, S. Bell, J. Irvine, and J. Banford, ‗Nodobo: Detailed
structure‘, Massachusetts Institute of Technology, 2011 mobile phone usage dataset‘, Unpublished paper, accessed at
[38] H. Sarvari, E. Abozinadah, A. Mbaziira, and D. Mccoy, ‗Constructing http://nodobo. com/papers/iet-el. pdf on, pp. 9–21, 2013.
and Analyzing Criminal Networks‘, 2014, pp. 84–91. [47] C. Kiss and Bichler Martin, ‗Identification of influencers—measuring
[39] J. Shetty and J. Adibi, ‗Discovering important users through graph influence in customer networks‘, Decision Support Systems, vol. 46,
entropy the case of enron email database‘, in Proceedings of the 3rd no. 1, pp. 233–253, 2008.
international workshop on Link discovery, 2005, pp. 74–81. [48] O. Hinz, B. Skiera, C. Barrot, and J. U. Becker, ‗Seeding strategies
[40] F. Tutzauer, ‗Entropy as a measure of centrality in networks for viral marketing: An empirical comparison‘, Journal of Marketing,
characterized by path-transfer flow‘, Social Networks, vol. 29, no. 2, vol. 75, no. 6, pp. 55–71, 2011.
pp. 249–265, May 2007. [49] S. Catanese, E. Ferrara, and G. Fiumara, ‗Forensic analysis of phone
[41] D. Ortiz-Arroyo and D. M. A. Hussain, ‗An Information Theory call networks‘, Social Network Analysis and Mining, vol. 3, no. 1, pp.
Approach to Identify Sets of Key Players‘, in Intelligence and 15–33, Mar. 2013.
Security Informatics, vol. 5376, D. Ortiz-Arroyo, H. L. Larsen, D. D. [50] E. Ferrara, P. De Meo, S. Catanese, and G. Fiumara, ‗Detecting
Zeng, D. Hicks, and G. Wagner, Eds. Berlin, Heidelberg: Springer criminal organizations in mobile phone networks‘, Expert Systems
Berlin Heidelberg, 2008, pp. 15–26 with Applications, vol. 41, no. 13, pp. 5733–5750, 2014.
219