Project Description

Consider the idea that leaderless terrorist networks can be more efficiently defeated by another network. With this idea in mind, read the articles posted in this topic and find similar ones in the GCU digital library. Construct a model of a terrorist network (actual or abstract) and describe its characteristic. Explain how, in your opinion an adversarial relationship between two networks in general can be defined, modeled, and describe. Adapt the abstract adversarial concept to devising a network tasked with defeating an actual terrorist network. Specifically:

Topic Explanation

“It takes a network to defeat a network has become a new U.S. Army counterinsurgency doctrine in that intelligence agencies have applied network analysis to identify and map links among terrorists and insurgent organizations, as well as in that the intelligence forces operate as a network to destroy the enemies” (Social Network Analysis: Methods and Examples, 151).

Since being credited with the discovery of Osama bin Laden’s hideout in 2013, social network analysis has been used to combat terrorism. Studying terrorism networks allows for the identification of the best responses and tactics in attempting to disrupt the cell. “As terrorist organizations often operate in a loose network form in which individual terrorists cooperate and collaborate with each other to carry out attacks (Sageman, 2004), we gain valuable knowledge by studying the structural properties of those networks” (Social Network Analysis: Methods and Examples, 151).

Data Explanation

The data used in this report is the Noordin Top Terrorist Network Data from the Association of Religion Data Archives.

“The Noordin Top Terrorist Network Data were drawn primarily from”Terrorism in Indonesia: Noordin’s Networks," a publication of the International Crisis Group, and include relational data on 79 individuals discussed in that publication. The data set includes information on these individuals’ affiliations with terrorist/insurgent organizations, educational institutions, businesses, and religious institutions. It also outlines which individuals are classmates, kin, friends, and co-religionists, and it details which individuals provided logistical support or participated in training events, terrorist operations, and meetings." -http://www.thearda.com/Archive/Files/Descriptions/TERRNET.asp

Principal Investigators:

  1. Nancy Roberts, Naval Postgraduate School
  2. Sean Everton, Naval Postgraduate School

For the purposes of this report, the terrorists will only be linked with terrorist/insurgent organizations.

Data Frame Representation

The data is read in:

raw.data = read.csv("Noordin Top Terrorist Network Data.csv", row.names = 1)

The first few rows and columns of the data are shown. There are 32 columns (organizations) total:

library(knitr)

kable(head(raw.data)[,1:11])
ORGAN1 ORGAN2 ORGAN3 ORGAN4 ORGAN5 ORGAN6 ORGAN7 ORGAN8 ORGAN9 ORGAN10 ORGAN11
Abdul Malik 0 0 0 0 1 0 0 0 0 0 0
Abdul Rauf 0 0 0 0 1 0 0 0 0 0 0
Abdul Rohim 0 0 0 0 0 0 0 0 0 0 0
Abdullah Sunata 0 0 0 0 0 0 0 0 0 0 0
Abdullah Sungkar 0 0 0 0 1 0 0 0 0 0 0
Abu Bakar Ba’asyir 0 0 0 0 1 0 0 0 0 1 0

Network Representation

This is a visualization of the network:

library(igraph)

net = graph_from_incidence_matrix(raw.data)

plot(net, vertex.label.color = "black", vertex.label.cex = 0.6,
     vertex.label = c(paste0("T",1:table(V(net)$type)[[1]]), paste0("O",1:table(V(net)$type)[[2]])),
     vertex.color = c("#7F8C8D", "#E74C3C")[1 + V(net)$type])
legend(x = -2.5, y = -1.0, c("Terrorist", "Organization"), pt.bg = c("#7F8C8D", "#E74C3C"), bty = "n",
       pch = 21, pt.cex = 2)

All the vertices that do not contain an edge are removed:

net = delete.vertices(net, which(degree(net) < 1))
plot(net, vertex.label.color = "black", vertex.label.cex = 0.6,
     vertex.label = c(paste0("T",1:table(V(net)$type)[[1]]), paste0("O",1:table(V(net)$type)[[2]])),
     vertex.color = c("#7F8C8D", "#E74C3C")[1 + V(net)$type])
legend(x = -2.5, y = -1.0, c("Terrorist", "Organization"), pt.bg = c("#7F8C8D", "#E74C3C"), bty = "n",
       pch = 21, pt.cex = 2)

Network Analysis

Describing the Terrorist Network

In order to analyze the terrorist network, two bipartite projections are made. One projection is a network of terrorists that are connected if they participate in the same terrorist organization or not, and the other is a network of organizations that are connected if they share at least one similar terrorist member. The two networks are shown below:

par(mfrow = c(1,2))
net.bp = bipartite.projection(net)

plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = "#7F8C8D")
legend(x = -1.8, y = -1.0, "Terrorist", pt.bg = "#7F8C8D", bty = "n", pch = 21, pt.cex = 2)

plot(net.bp$proj2, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("O",1:table(V(net)$type)[[2]]), vertex.color = "#E74C3C")
legend(x = -1.8, y = -1.0, "Organization", pt.bg = "#E74C3C", bty = "n", pch = 21, pt.cex = 2)

From this point on I will refer to these networks as the “terrorist projection” and the “organization projection”.

Degree

“Members with high degrees are likely to be the leaders of their local networks” (Social Network Analysis: Methods and Examples, 153).

These are the nodes with the highest degrees in the terrorist projection:

degree(net.bp$proj1)[degree(net.bp$proj1) >= 20]
##            Abdul Rauf      Abdullah Sungkar            Agus Ahmad 
##                    20                    32                    20 
##     Ahmad Rofiq Ridho      Asmar Latin Sani          Azhari Husin 
##                    20                    20                    20 
##               Chandra               Cholily              Dulmatin 
##                    32                    20                    20 
## Fathurrahman al-Ghozi          Hari Kuncoro          Imam Samudra 
##                    21                    29                    29 
##          Irun Hidayat                Ismail                 Jabir 
##                    20                    20                    20 
##         Mohamed Ihsan               Muchtar     Sardona Siliwangi 
##                    20                    20                    20 
##        Subur Sugiarto                 Tohir            Toni Togar 
##                    20                    20                    20 
##                 Ubeid                  Umar            Umar Patek 
##                    20                    20                    20

The terrorists with the highest degree are Abdullah Sungkar and Chandra. These terrorists are most likely to be the core/leaders of the terrorist networks. They can be seen highlighted below:

vcol = rep("#7F8C8D", vcount(net.bp$proj1))
vcol[V(net.bp$proj1)$name == "Abdullah Sungkar" | V(net.bp$proj1)$name == "Chandra"] = "#F4D03F"

plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Terrorist", "Likely Leaders"), pt.bg = c("#7F8C8D", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

Betweeness

“Individuals with high betweenness are usually the contact persons among several terrorist groups and play important roles in coordinating terrorist attacks” (Social Network Analysis: Methods and Examples, 153).

These are the nodes with the highest betweeness in the terrorist projection:

betweenness(net.bp$proj1)[betweenness(net.bp$proj1) >= 300]
## Abu Bakar Ba'asyir       Hari Kuncoro 
##           392.1891           432.4241

These terrorists are most likely to play important roles in coordinating the different organizations. They can be seen highlighted below:

vcol = rep("#7F8C8D", vcount(net.bp$proj1))
vcol[V(net.bp$proj1)$name == "Abu Bakar Ba'asyir" | V(net.bp$proj1)$name == "Hari Kuncoro"] = "#F4D03F"

plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Terrorist", "Likely Organizers"), pt.bg = c("#7F8C8D", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

Outliers

They also found that outliers with particularly low closeness can be the true leaders of local networks, only appearing to be marginal while directing the whole network from behind the scenes to evade the attention of intelligence agencies.

These are the nodes with a degree of one in the terrorist projection:

degree(net.bp$proj1)[degree(net.bp$proj1) == 1]
## Ahmad Sayid Maulana        Imam Bukhori             Qotadah 
##                   1                   1                   1

The terrorists that might be the true leaders, who control the organization from the outside, are likely to be those with a low degree. They can be seen highlighted below:

vcol = rep("#7F8C8D", vcount(net.bp$proj1))
vcol[V(net.bp$proj1)$name == "Ahmad Sayid Maulana" | V(net.bp$proj1)$name == "Imam Bukhori" |
       V(net.bp$proj1)$name == "Qotadah"] = "#F4D03F"

plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Terrorist", "Likely Covert Leaders"), pt.bg = c("#7F8C8D", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

Describing the Counter-terrorist Network

“Destabilization strategies targeting a terrorist network can be roughly categorized into two types: one aiming at isolating or decapitating particular actors or leaders, and the other aiming at destroying the network’s infrastructure, such as weapons or training facilities and transportation or communication channels” (Social Network Analysis: Methods and Examples, 155).

Removing Key Actors

Key actors in a terrorist organization are members that would severely damage the communication/infrastructure of the terrorist network if they are removed from it. These key actors could be leaders, coordinators, or some other crucial role to the network itself.

In the previous section, Describing the Terrorist Network, these key actors in the terrorist network were all identified. Removing these key actors, via disabling communication, discrediting, or even assassinating, would dismantle the command-and-control infrastructure of the network and fully disconnect cells from crucial operating information or resource flow.

The network visualizations previously shown that identified the key actors in the terrorist projection can again be seen below:

par(mfrow=c(1,3))

vcol = rep("#7F8C8D", vcount(net.bp$proj1))
vcol[V(net.bp$proj1)$name == "Abdullah Sungkar" | V(net.bp$proj1)$name == "Chandra"] = "#F4D03F"
plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Terrorist", "Likely Leaders"), pt.bg = c("#7F8C8D", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

vcol = rep("#7F8C8D", vcount(net.bp$proj1))
vcol[V(net.bp$proj1)$name == "Abu Bakar Ba'asyir" | V(net.bp$proj1)$name == "Hari Kuncoro"] = "#F4D03F"
plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Terrorist", "Likely Organizers"), pt.bg = c("#7F8C8D", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

vcol = rep("#7F8C8D", vcount(net.bp$proj1))
vcol[V(net.bp$proj1)$name == "Ahmad Sayid Maulana" | V(net.bp$proj1)$name == "Imam Bukhori" |
       V(net.bp$proj1)$name == "Qotadah"] = "#F4D03F"
plot(net.bp$proj1, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("T",1:table(V(net)$type)[[1]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Terrorist", "Likely Covert Leaders"), pt.bg = c("#7F8C8D", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

Disabling Key Facilities

Terrorist networks are normally distributed into quasi-independent cells that commands are distributed to. These cells are normally small but are self-sustaining. They are loosely connected to each other, so finding the cells/organizations that hold the most significant role in a terrorist network is valuable in disabling the communication between cells.

The network projection of the full terrorist organization connects smaller independent organizations based on whether they have the same terrorist members or not. The organization in this network with the highest betweeness is likely to be the core organization in the network that handles most of the communication in the overall terrorist network between every organization.

These are the nodes in the organization projection that have the highest betweeness:

betweenness(net.bp$proj2)[betweenness(net.bp$proj2) >= 50]
##  ORGAN5 ORGAN10 ORGAN18 ORGAN20 
##     160      92     129      73

These organizations are highlighted in the plot below:

vcol = rep("#E74C3C", vcount(net.bp$proj1))
vcol[V(net.bp$proj2)$name == "ORGAN5" | V(net.bp$proj2)$name == "ORGAN10" |
       V(net.bp$proj2)$name == "ORGAN18" | V(net.bp$proj2)$name == "ORGAN20"] = "#F4D03F"

plot(net.bp$proj2, vertex.label.color = "black", vertex.label.cex = 0.5,
     vertex.label = paste0("O",1:table(V(net)$type)[[2]]), vertex.color = vcol)
legend(x = -1.8, y = -1.0, c("Organization", "Communication Hub"), pt.bg = c("#E74C3C", "#F4D03F"),
       bty = "n", pch = 21, pt.cex = 2)

Disabling or destroying these organizations in some way would greatly hinder the terrorist network.

References

(n.d.). Retrieved February 24, 2018, from http://www.thearda.com/Archive/Files/Descriptions/TERRNET.asp

Yang, S., Zhang, L., & Keller, F. B. (2017). Social network analysis: methods and examples. Los Angeles: Sage.