Detection of Terrorism-related Twier Communities using
Ilias Gialampoukidis, George Kalpakis, eodora Tsikrika, Symeon Papadopoulos, Stefanos
Vrochidis and Ioannis Kompatsiaris
Information Technologies Institute
Centre for Research and Technology Hellas
essaloniki, Greece 57001
Social media are widely used among terrorists to communicate and
disseminate their activities. User-to-user interaction (e.g. mentions,
follows) leads to the formation of complex networks, with topology
that reveals key-players and key-communities in the terrorism
domain. Both the administrators of social media platforms and Law
Enforcement Agencies seek to identify not only single users but
groups of terrorism-related users so that they can reduce the impact
of their information exchange eorts. To this end, we propose a
novel framework that combines community detection with key-
player identication to retrieve communities of terrorism-related
social media users. Experiments show that most of the members of
each retrieved key-community are already suspended by Twier,
violating its terms, and are hence associated with terrorism-oriented
content with high probability.
•Information systems →Information retrieval; Test collec-
Web searching and information discovery; Multimedia and
•Human-centered computing →Social
networking sites; •Networks →Social media networks;
Social Network Analysis, key-player identication, community
detection, terrorism-oriented social media mining, Twier
ACM Reference format:
Ilias Gialampoukidis, George Kalpakis, eodora Tsikrika, Symeon Pa-
padopoulos, Stefanos Vrochidis and Ioannis Kompatsiaris. 2017. Detec-
tion of Terrorism-related Twier Communities using Centrality Scores. In
Proceedings of MFSec’17, Bucharest, Romania, June 06, 2017, 5 pages.
e rapid growth of the Internet has resulted in modern forms
of communication and exchange of information, realized mainly
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for components of this work owned by others than the
author(s) must be honored. Abstracting with credit is permied. To copy otherwise, or
republish, to post on servers or to redistribute to lists, requires prior specic permission
and/or a fee. Request permissions from firstname.lastname@example.org.
MFSec’17, Bucharest, Romania
2017 Copyright held by the owner/author(s). Publication rights licensed to ACM.
978-1-4503-5034-1/17/06. . . $15.00
through the use of social media networking platforms (e.g. Twier,
Facebook, etc.), which have dominated the online world during
the past few years. Social media networks have made possible
the communication among people across nationalities, religions,
cultures or residences; however, their great power and reach has
become an aractive feature for their use by terrorist and extremist
organizations for disseminating their propaganda, recruiting and
radicalizing new members, raising funds, organizing operations,
and publishing information and instructions exploited by lone-wolf
terrorists when preparing and commiing acts of terror [27–29].
Due to its nature that permits the inexpensive communication of
multimedia messages (i.e. tweets) to users worldwide, Twier has
been used primarily for promoting and spreading their propaganda
typically using a top-down approach, with a core group of members
spreading the group’s messages, which are then re-shared by other
aliated accounts. Both the administrators of the social media
networking platform itself (Twier), on the one hand, and the
Law Enforcement Agencies (LEAs), on the other, are interested in
monitoring terrorism-related activities taking place through the
platform. In the former case, the goal is to detect material that
violates the platform’s terms and conditions regarding extremist
content, while in the laer case such information may be very useful
in investigations for prosecuting the perpetrators of terrorist aacks.
In both cases, it is of vital signicance to detect the communities
in the social networks and their most prominent users (i.e. key
players) who disseminate terrorism-related information, so as to
prevent terrorist groups from spreading their propaganda (to the
extent possible), by shuing down accounts who are found to play
a central role in this information exchange.
Over the past two decades, several research eorts have dis-
cussed the network structure of terrorist organizations. One of the
early eorts examined the network structure of the 9/11 hijackers
along with their accomplices and detected the ring leaders of the
terrorist aacks based on their social associations [
]. Later work
focused on using social network analysis for examining the basic
characteristics of terrorist groups or organizations [
]. More re-
cent research has examined the survival mechanisms of the Global
Sala Jihad (GSJ) terrorist network, even aer being severely dam-
aged by the authorities, by analyzing its network structure and
]. In addition, several works have been conducted for
studying the use of social media, and especially Twier, by terrorist
organizations. Specically, a work has examined the signicant role
of Twier in facilitating terrorists to execute their aack in Mumbai
(November 2008), by monitoring and exploiting situational infor-
mation which was broadcast through Twier [
]. More recent
MFSec’17, June 06, 2017, Bucharest, Romania I. Gialampoukidis et al.
research has studied the Islamic State’s (IS) strategy for communi-
cating their propaganda for radicalizing and recruiting Twier users
]. Furthermore, the signicant role played by feeder accounts of
terrorist organizations for exchanging information from the Syria
insurgency zone is pointed out in [
]. Key player identication in
complex networks, on the other hand, has been mainly addressed
through the use of dierent centrality measures; e.g. recent work
] has used several centrality measures to rank terrorism-related
Twier accounts based on their location in the network and the
topology of the network of user-to-user mentions.
is work aims at identifying groups of terrorism-related users
exchanging information through social media platforms by detect-
ing the key players of a social media network and the interrelated
communities of users interacting with them. To this end, we extend
the approach of [
] and propose a hybrid framework which rst
retrieves the key network players and then enriches the retrieved
results by adding the members of a user’s detected community
based on the combination of centrality scores with community
detection algorithms. ese centrality measures, which aim to iden-
tify key-players in the terrorism domain, are estimated on social
media networks based on user mentions and are compared with
other popularity measures (i.e. number of followers, number of
friends) used for identifying very important users within the struc-
ture of these networks. is work also presents a case study on a
social media network formed by Twier accounts based on a set of
terrorism-related Arabic keywords provided by LEAs and domain
experts, for demonstrating the performance of our proposed frame-
work based on evidence related to the suspension of the majority
of the retrieved Twier accounts.
2 KEY TERRORISM COMMUNITY
In this work, entropy-based centrality measures are exploited to
rst retrieve a list of key-players and then a community detection
algorithm to enrich the initial set of results. Our framework is
presented in Figure 1, where keyword-based search provides a set of
social media posts. Based on this, a network of mentions is created,
using the user-to-user interactions contained in the corresponding
posts. In the resulting network of users, each user is represented by
a node and a link between two users
exists if user
or is mentioned by user nk.
On the network of mentions, we use entropy-based centrality
measures to, rst, identify key-players [
] and we then extend the
method by associating key-players with their community.
2.1 Centrality-based key player identication
We denote by
the network of mentions with
links. e network is unweighted and undirected
capturing only the user-to-user interactions in Twier or any other
social media domain. e degree of a node
is denoted by
and is equal to the number of its adjacent links. e degree is
normalized to dene the degree centrality as follows :
e degree simply counts the number of nodes and is not aected
by the position of a hub in the network. However, the betweenness
] of a node
is based on the number of paths
that pass through node
, divided by the
number of all paths
, summed over all
pairs of nodes (ni,nj)and normalized by its maximum value:
Nodes with high betweenness centrality are very important for the
communication in a network [
] , due to the fact that their removal
strongly aects the network connectivity and robustness. Other
centrality measures have also been proposed, based on the mutual
distances of all nodes (closeness centrality) [
], on the inuence of
a node (eigenvector centrality) [
], or motivated by the importance
of a Web page (PageRank) .
In the context of this work, we propose the use of entropy-based
centrality measures, such as the Mapping Entropy (ME) and the
Mapping Entropy Betweenness (MEB), taking also into account the
of a node
.Mapping Entropy centrality [
is dened as a function of the degree centrality:
whereas Mapping Entropy Betweenness centrality [
] is dened as
a function of betweenness centrality:
Intuitively, to interpret Equations (3) and (4), one may think of
a random walker on the network, standing at node
, who picks
his/her next step with probability
). en, the weight
) is interpreted as the Shannon information of
the event that the random walker picked node
, and is summed
over all neighbors of node
. ese two measures consider the
information that is communicated through nodes who act as a hub
(bridge), i.e. those with high values of degree (betweenness) cen-
trality between any two members. In particular, the MEB centrality
considers the betweenness centrality of a node and exploits local
information from its neighborhood; hence, high MEB values indi-
cate that a particular node can act as a bridge for disseminating
information, even if their degree centrality is low .
In the following, we combine the key-player identication meth-
ods with community detection approaches that are able to cluster
the network into communities of densely connected user accounts.
2.2 Community detection around key players
In parallel to the key-player identication, a community detection
algorithm is used to divide the network into groups of users (com-
munities). e top-ranked key-player is used to enrich the retrieved
results, which is achieved by searching for the community where
the key-player belongs to.
Community detection in complex networks aims to identify
groups of nodes that are more densely connected to each other
within a group than to the rest of the network outside of the group
]. e groups are communities of users in the social media do-
main, sharing a common property or playing similar roles within
the network [
]. Community structure is very popular in many
elds, including sociology and biology [
], as well as computer
Detection of Terrorism-related Twier Communities using Centrality Scores MFSec’17, June 06, 2017, Bucharest, Romania
Search by Keyword Network of
Mentions Mapping Entropy
Figure 1: Key terrorism-related community detection on the network of Twitter mentions.
], and in any domain where systems or items admit a
network representation. Detecting communities in complex net-
works is oen viewed as a graph partitioning problem, where all
nodes are assigned to a community, but density-based approaches
leave out noise, i.e. do not assign all nodes to communities. In our
experiments, we shall present and compare both approaches.
Several community detection algorithms have been proposed
]). e network is partitioned into
communities using either the maximization of modularity [
the minimization of codelength [
] or density-based approaches
]. We present in the experiments the key-community, dened as
the community that the key-player belongs to, as provided by the
algorithms FastGreedy [
], Walktrap [
], Infomap [
] and DBSCAN*-Martingale [
]. e most popular methods
are those aiming at the maximization of modularity, dened as [
is the fraction of links between a node in community
and a node in community
is the fraction of links between
two members of the community
number of communities. We adopt the modularity maximization
community detection approach as a fast and scalable approach
that admits hierarchical and iterative methods [
] to maximize
the objective function of Equation 5. Assuming the key-player is
a member of the
-th community, our framework returns all its
nk1,nk2, . . . ,nkl
, all of which are marked as the nal list
of accounts with suspicious activity.
We evaluate our framework in a network consisting of terrorism-
related Twier accounts formed based on user mentions.
As ground-truth we make use of information from Twier, which
marks user accounts as suspended, given that the suspension pro-
cess is applied when an account violates Twier rules by exhibit-
ing abusive behavior, including posting content related to violent
threats and hate speech (Twier has suspended 360,000 terrorism-
related accounts from mid-2015 until August 2016
). Our data were
collected by executing queries on the Twier API
based on a set
of ve Arabic keywords related to terrorist propaganda. ese
keywords were provided by LEAs and domain experts and are re-
lated to the Caliphate State, its news, publications, and photos from
the Caliphate area. e collected dataset consists of 9,528 Twier
posts by 4,400 users. e top-100 user accounts are retrieved in the
key-player identication step using the ranking methods of Table 1
and are then combined with the community detection approaches
of Table 2. e evaluation is performed by assessing whether these
accounts are suspended, active or no longer exist (i.e. accounts
which have been temporarily or permanently deactivated).
e rst part of our framework evaluates several centrality mea-
sures, including the proposed Mapping Entropy and Mapping En-
tropy Betweenness, as well as popularity measures, such as the
number of friends and followers, in terms of their ability to re-
trieve suspended users. e results in Table 1 indicate that the
entropy-based centralities ME and MEB are able to retrieve the rst
suspended user at position 16, while PageRank follows at position
19. Other centrality and popularity measures, such as closeness,
eigenvector and number of followers do not nd any suspended
user at the top-100 positions of their retrieved users. We observe
that the network is very spread with many bridges and a diam-
eter equal to 27, so key-players are expected to be positioned in
MFSec’17, June 06, 2017, Bucharest, Romania I. Gialampoukidis et al.
Table 1: Comparison among several ranking methods
Ranking Method Position Reciprocal Rank
Degree centrality 20 5.00%
Betweenness centrality 35 2.86%
Closeness centrality >100 <1.00%
Eigenvector centrality >100 <1.00%
Num of followers >100 <1.00%
Num of friends 31 3.23%
PageRank 19 5.26%
Mapping Entropy 16 6.25%
Mapping Entropy Betweenness 16 6.25%
K=1 K=2 K=3
K=5 K=10 Largest connected component
Figure 2: First, second, third, h and tenth order neighbor-
hoods of the suspended user and the largest component.
between many pairs of nodes in the network, exploiting also their
neighborhood’s high betweenness centrality.
-th order neighborhood
is the set of all
nodes that are reachable from
1 intermediate nodes:
d(n,nj) ≤ K}
is the network
distance of any two nodes. In Figure 2 we show the rst
order neighborhoods of the rst suspended user and the largest
connected component. Although the ME and MEB centralities both
retrieve a suspended user at rank 16, the user does not correspond
to the same Twier account. In fact, the Twier user at the 16
position of ME centrality leads to a disconnected component of
two users, where one of them is suspended and the other is not.
However, the neighborhood of the suspended user (Figure 2) from
the MEB centrality is part of the largest connected component of
the network with 1,334 accounts. erefore, we proceed to the next
step by considering the MEB centrality measure and not ME.
Given the rst identied suspended user in the MEB ranking,
we explore the community where the user belongs to. e re-
sults are reported in Table 2, along with the community size per
community detection method. We observe that in all cases exam-
ined, the majority of accounts are already suspended and some of
Table 2: Comparison of community detection methods
FastGreedy 58 4 (6.9%)
Walktrap 33 3 (9.1%)
Louvain 58 4 (6.9%)
23 2 (8.7%)
Figure 3: A sample set of images uploaded by key-players
with militaristic or nationalistic content. Faces are redacted
so as to avoid the inclusion of sensitive information.
them no longer exist. In particular, the modularity maximization
methods (FastGreedy, Louvain) are able to retrieve the largest com-
munities and thus more accounts with potentially illegal activity.
e percentage of suspended users is 82.76% for the modularity
maximization approaches and 78% for the Walktrap and DBSCAN*-
Martingale, indicating a marginal advantage for the former. e
community provided by Infomap is very small, compared to the
other community sizes, but still the number of active accounts (not
yet suspended) is only 20%. Figure 3 depicts sample content from
such active accounts that have not been marked as suspended by
Twier. One may note that their content is military-themed or
extremist, indicating potentially suspicious user activity even in
We proposed a hybrid model that combines MEB centrality and com-
munity detection that retrieves groups of social media user accounts
that are key-players in the terrorism domain. We found that central-
ity measures on the network of mentions perform beer than other
popularity measures (number of followers or friends) in nding
key-players in the terrorism domain. Given a terrorism-related user,
his/her network community reveals a group of additional terrorism-
related users, exploiting the outcome of a community detection
method, with modularity maximization methods outperforming
density-based and other methods.
is work was supported by the project TENSOR (H2020-700024),
funded by the European Commission.
Detection of Terrorism-related Twier Communities using Centrality Scores MFSec’17, June 06, 2017, Bucharest, Romania
Ala Berzinji, Lisa Kaati, and Ahmed Rezine. 2012. Detecting key players in
terrorist networks. In Intelligence and Security Informatics Conference (EISIC),
2012 European. IEEE, 297–302.
Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambioe, and Etienne Lefeb-
vre. 2008. Fast unfolding of communities in large networks. Journal of statistical
mechanics: theory and experiment 2008, 10 (2008), P10008.
Ludvig Bohlin, Daniel Edler, Andrea Lancichinei, and Martin Rosvall. 2014.
Community detection and visualization of networks with the map equation
framework. In Measuring Scholarly Impact. Springer, 3–34.
Phillip Bonacich and Paulee Lloyd. 2001. Eigenvector-like measures of centrality
for asymmetric relations. Social networks 23, 3 (2001), 191.
Sergey Brin and Lawrence Page. 2012. Reprint of: e anatomy of a large-scale
hypertextual web search engine. Computer networks 56, 18 (2012), 3825–3833.
Akemi Takeoka Chateld, Christopher G Reddick, and Uuf Brajawidagda. 2015.
Tweeting propaganda, radicalization and recruitment: Islamic state supporters
multi-sided twier networks. In Proceedings of the 16th Annual International
Conference on Digital Government Research. ACM, 239–249.
Aaron Clauset, Mark EJ Newman, and Cristopher Moore. 2004. Finding commu-
nity structure in very large networks. Physical review E 70, 6 (2004), 066111.
Santo Fortunato. 2010. Community detection in graphs. Physics reports 486, 3
Linton C Freeman. 1978. Centrality in social networks conceptual clarication.
Social networks 1, 3 (1978), 215–239.
Ilias Gialampoukidis, George Kalpakis, eodora Tsikrika, Stefanos Vrochidis,
and Ioannis Kompatsiaris. 2016. Key player identication in terrorism-related
social media networks using centrality measures. In European Intelligence and
Security Informatics Conference (EISIC 2016), August. 17–19.
Ilias Gialampoukidis, eodora Tsikrika, Stefanos Vrochidis, and Ioannis Kom-
patsiaris. 2016. Community detection in complex networks based on DBSCAN*
and a Martingale process. In Semantic and Social Media Adaptation and Personal-
ization (SMAP), 2016 11th International Workshop on. IEEE, 1–6.
Michelle Girvan and Mark EJ Newman. 2002. Community structure in social
and biological networks. Proceedings of the national academy of sciences 99, 12
Steve Harenberg, Gonzalo Bello, L Gjeltema, StephenRanshous, Jitendra Harlalka,
Ramona Seay, Kanchana Padmanabhan, and Nagiza Samatova. 2014. Community
detection in large-scale networks: a survey and empirical evaluation. Wiley
Interdisciplinary Reviews: Computational Statistics 6, 6 (2014), 426–439.
Jye Klausen. 2015. Tweeting the Jihad: Social media networks of Western
foreign ghters in Syria and Iraq. Studies in Conict & Terrorism 38, 1 (2015),
 Valdis Krebs. 2002. Uncloaking terrorist networks. First Monday 7, 4 (2002).
Fragkiskos D Malliaros and Michalis Vazirgiannis. 2013. Clustering and com-
munity detection in directed networks: A survey. Physics Reports 533, 4 (2013),
ME Newman and M Girvan. 2004. Finding and evaluating community structure
in networks. Physical Review E 69, 2 (2004), 026113.
Tingyuan Nie, Zheng Guo, Kun Zhao, and Zhe-Ming Lu. 2016. Using mapping
entropy to identify node centrality in complex networks. Physica A: Statistical
Mechanics and its Applications 453 (2016), 290–297.
Onook Oh, Manish Agrawal, and H Raghav Rao. 2011. Information control and
terrorism: Tracking the Mumbai terrorist aack through twier. Information
Systems Frontiers 13, 1 (2011), 33–43.
Symeon Papadopoulos, Yiannis Kompatsiaris, Athena Vakali, and Ploutarchos
Spyridonos. 2012. Community detection in social media. Data Mining and
Knowledge Discovery 24, 3 (2012), 515–554.
Pascal Pons and Mahieu Latapy. 2006. Computing communities in large net-
works using random walks. J. Graph Algorithms Appl. 10, 2 (2006), 191–218.
Jialun Qin, Jennifer J Xu, Daning Hu, Marc Sageman, and Hsinchun Chen. 2005.
Analyzing terrorist networks: A case study of the global sala jihad network. In
Intelligence and security informatics. Springer, 287–304.
Usha Nandini Raghavan, R
eka Albert, and Soundar Kumara. 2007. Near linear
time algorithm to detect community structures in large-scale networks. Physical
Review E 76, 3 (2007), 036106.
Martin Rosvall, Daniel Axelsson, and Carl T Bergstrom. 2009. e map equation.
e European Physical Journal Special Topics 178, 1 (2009), 13–23.
Martin Rosvall and Carl T Bergstrom. 2008. Maps of random walks on complex
networks reveal community structure. Proceedings of the National Academy of
Sciences 105, 4 (2008), 1118–1123.
Sudhir Saxena, K Santhanam, and Aparna Basu. 2004. Application of social
network analysis (SNA) to terrorist networks in Jammu & Kashmir. Strategic
Analysis 28, 1 (2004), 84–101.
Robin L ompson. 2011. Radicalization and the use of social media. Journal of
strategic security 4, 4 (2011), 167.
Robyn Torok. 2010. “Make A Bomb In Your Mums Kitchen”: Cyber Recruiting
And Socialisation of ‘White Moors’ and Home Grown Jihadists. (2010).
Ines Von Behr. 2013. Radicalisation in the digital era: e use of the Internet in
15 cases of terrorism and extremism. (2013).
Jie Xu, Daning Hu, and Hsinchun Chen. 2009. e dynamics of terrorist net-
works: Understanding the survival mechanisms of Global Sala Jihad. Journal
of Homeland Security and Emergency Management 6, 1 (2009), 1–15.