06.

Network
Centrality
Lecturer: Dr. Reem Essameldin Ebrahim

Introduction to Social
Based on CS224W Analysis of Networks Mining and Learning with Graphs: Stanford University
Copyright © Dr. Reem Essameldin 2023-2024
 Why/What is Network Centrality?

In this Lecture Different Centrality Measures

Topics to be covered are:

Copyright © Dr. Reem Essameldin 2023-2024

What is Centrality?
Centrality measures address the question: "Who is the most important or central
person in this network?" There are many answers to this question, depending on
what we mean by importance.
Centrality tells us which nodes are important in a
network based on the topological structure of the
network, such as:
• How influential a person is within a social
network?
• Which genes play a crucial role in regulating
systems and processes?
• Infrastructure networks: if the node is removed, it
would critically impede the functioning of the
network.

Copyright © Dr. Reem Essameldin 2023-2024

Why Network Centrality?
From the structure of a network, we can calculate various useful quantities
(measures) that capture particular features of the network topology. Finding out
which is the most central node is important, it could help with:

Disseminating information in the network faster.

Stopping epidemics.
Protecting the network from breaking.
Evenly distributing centrality among nodes.
Determining which nodes are most ‘central’?

Copyright © Dr. Reem Essameldin 2023-2024

Centrality Measures
One first needs to define measures for quantifying centrality, level of interactions
and similarity, among others. Centrality measures help to understanding social
networks or social graphs.
There are various measures that are used to
evaluate the centrality of the complex networks.
Each measure has its own definition of
‘importance’. We will look at some popular
ones.

Examples:
Degree Centrality (who you are connected to)
Betweenness Centrality (how you connect others)
Closeness Centrality (how fast you can reach others)
Eigenvector Centrality (who you are connected to)
PageRank Centrality

Copyright © Dr. Reem Essameldin 2023-2024

 Degree Centrality Assigns importance score based on no. of links.

Degree centrality is the simplest measure of node connectivity. It tells how many
direct connections to other nodes. It can be used for finding very connected
individuals, popular individuals, individuals who are likely to hold most
Centrality Measures

information or individuals who can quickly connect with the wider network.

Node degree is local, not global

Eigen centrality

Copyright © Dr. Reem Essameldin 2023-2024

 Degree Centrality Assigns importance score based on no. of links.

Degree centrality is the simplest measure of node connectivity. It tells how many
direct connections to other nodes. It can be used for finding very connected
individuals, popular individuals, individuals who are likely to hold most
Centrality Measures

information or individuals who can quickly connect with the wider network.

Normalized Degree Centrality

The calculated degree of a node divided by
maximum possible degree of the network, (𝑵 − 𝟏):

𝑛𝑜𝑟𝑚𝑎𝑙𝑖𝑧𝑒𝑑 𝑑𝑒𝑔𝑟𝑒𝑒 𝑐𝑒𝑛𝑡𝑟𝑎𝑙𝑖𝑡𝑦=𝑑𝑒𝑔𝑟𝑒𝑒𝑁−1

𝑜𝑓 𝑛𝑜𝑑𝑒 𝑖

Where, 𝑵 is total number of nodes in the network.

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the degree centrality of each node.

Solution:
1

0 2

6 7

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the degree centrality of each node.

Solution:
1

0 2

6 7

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the normalized degree centrality of each node.

Solution:

? ?

? ? ?

? ?

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the normalized degree centrality of each node.

Solution:
2
= 0.33
7−1
? ? .3 .3

? ? ? .5 .3 .5

? 3 ? .3 .3
= 0.5
7−1

Copyright © Dr. Reem Essameldin 2023-2024

 Eigenvector Centrality
Having more friends does not by itself guarantee that someone is more
important. Having more important friends provides a stronger signal.
Eigenvector centrality incorporates the importance of the neighbors.
Centrality Measures

Like degree centrality, eigenvector centrality

measures a node’s influence based on the
number of links it has to other nodes in the
network. Eigenvector centrality then goes a
step further by also taking into account how
well connected a node is, and how many links
their connections have, and so on through the
network.
Note that: Eigenvector centrality is a good ‘all-round’ SNA
score, handy for understanding human social networks, but also
for understanding networks like malware propagation.

Copyright © Dr. Reem Essameldin 2023-2024

 Eigenvector Centrality
Having more friends does not by itself guarantee that someone is more
important. Having more important friends provides a stronger signal.
Eigenvector centrality incorporates the importance of the neighbors.
Centrality Measures

The Mathematical Definition

Let us assume the eigenvector centrality of a node is 𝐶𝑒 𝑣𝑖 .
We would like 𝐶𝑒 𝑣𝑖 to be higher when important neighbors
(node 𝑣𝐽 with higher 𝐶𝑒 𝑣𝑖 point to the node. For incoming
neighbors 𝐴𝑗,𝑖 =1. We can assume that 𝑪𝒆 𝒗𝒊 centrality is the
summation of its neighbor’s centralities:
𝑛

𝐶𝑒 𝑣𝑖 = ෍ 𝐴𝑗,𝑖 𝐶𝑒 𝑣𝑗
𝑗=1

Copyright © Dr. Reem Essameldin 2023-2024

 Example
For the given graph, find the eigenvector centrality of each node.

The adjacency matrix A The Degrees of various Matrix vector V

A C of the given graph nodes are representation

1 2 3
B

D E

Now, mathematically the Eigen Vector Centrality is calculated as below:

4

Copyright © Dr. Reem Essameldin 2023-2024

 Betweenness Centrality How fast spread information

Betweenness centrality measures the number of times a node lies on the

shortest path between other nodes. This measure shows which nodes are
‘bridges’ between nodes in a network. It is used for finding the individuals who
Centrality Measures

influence the flow around a system.

The Mathematical Definition

It does this by identifying all the shortest paths
and then counting how many times each node
falls on one:
𝜎𝑠𝑡 (𝑣𝑖 )
𝐶𝑏 𝑣𝑖 = ෍
𝜎𝑠𝑡
𝑠≠𝑡≠𝑣
Where, 𝝈𝒔𝒕 the number of shortest paths from
vertex 𝒔 to 𝒕. 𝝈𝒔𝒕 (𝒗𝒊 ) the number of shortest
paths from s to t that passthrough 𝒗𝒊 .

Copyright © Dr. Reem Essameldin 2023-2024

 Betweenness Centrality How fast spread information

Betweenness centrality is useful for analyzing communication dynamics. A high

betweenness count could indicate that someone holds authority over disparate
clusters in a network, or just that they are on the periphery of both clusters.
Centrality Measures

Normalizing betweenness centrality

In the best case, node 𝒗𝒊 is on all shortest paths
𝜎𝑠𝑡 (𝑣𝑖 )
from 𝒔 to 𝒕, hence, =1
𝜎𝑠𝑡

𝐶𝑏 𝑣𝑖 = ෍ 1 = (𝑛 − 1)(𝑛 − 2)
𝑠≠𝑡≠𝑣

Therefore, the maximum value is,

𝐶 𝑏 𝑣𝑖
𝐶𝑏 𝑛𝑜𝑟𝑚 𝑣𝑖 =
(𝑛−1)(𝑛−2) /2

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the betweenness centrality of each node and its normalization.

Solution:

E D

B C

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the betweenness centrality of each node and its normalization.

Solution:
First, find the shortest path from each node to every node
E D and intermediate node in that path.

B C

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the betweenness centrality of each node and its normalization.

Solution:
From Table, node A is not an intermediate node in any
E D path. Thus, the betweenness centrality of node A is 0.
𝐶𝑏 𝐴 = 0

𝜎𝑠𝑡 (𝐵) 1 + 0.5

𝐶𝑏 𝐵 = ෍ = = 0.125
𝜎𝑠𝑡 12
𝑠≠𝑡≠𝐵
A
𝜎𝑠𝑡 = total shortest path =12

B C

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the betweenness centrality of each node and its normalization.

Solution:

E D

B C

What if it’s not so important to have many direct friends, or to

be ‘between ’others but one still wants to be in the ‘middle ’of
things, not too far from the center?
Copyright © Dr. Reem Essameldin 2023-2024
 Closeness Centrality How effectively control diffusion

Closeness centrality scores each node based on their ‘closeness’ to all other
nodes in the network. It is used for finding the individuals who are best placed to
influence the entire network most quickly(i.e. help find good ‘broadcasters’).
Centrality Measures

What may be more useful is using closeness to find influencers in a single cluster.

This measure calculates the shortest paths

between all nodes, then assigns each node a
score based on its sum of shortest paths.
Closeness is based on the length of the
average shortest path between a vertex and all
vertices in the graph.

Note that: In a highly-connected network, you will often

find all nodes have a similar score.

Copyright © Dr. Reem Essameldin 2023-2024

 Closeness Centrality How effectively control diffusion

The intuition is that influential and central nodes can quickly reach other nodes.
These nodes should have a smaller average shortest path length to other nodes.
It depends on inverse distance to other vertices.
Centrality Measures

The Mathematical Definition

It does this by identifying all the shortest paths
and then counting how many times each node
falls on one:
𝑁

𝐶𝑐 𝑖 = [෍ 𝑑(𝑖, 𝑗)]−1
𝑗=1

Normalized closeness centrality:

𝐶𝑐 ′ (𝑖) = 𝐶𝑐 𝑖 ∗ (𝑁 − 1)

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the closeness centrality of each node and its normalization.

Solution:

A B C D E

Copyright © Dr. Reem Essameldin 2023-2024

 Test Yourself
For the given graph, find the closeness centrality of each node and its normalization.

Solution:

′ σ𝑁
𝑗=1 𝑑(𝐴,𝑗)
1 + 2 + 3 + 4 −1
𝐶𝑐 𝐴 = [ ] −1 = [ ] = 0.4
𝑁−1 4

A B C D E .4 .57 .67 .57 .4

Copyright © Dr. Reem Essameldin 2023-2024

Copyright © Dr. Reem Essameldin 2023-2024
Copyright © Dr. Reem Essameldin 2023-2024
In the follower relationship network, if a user has a superior
degree centrality, a higher read probability would be
associated with his tweets, and there would be a higher
opportunity for his tweets to go viral. If a user’s closeness
centrality is high, the user's aptitude to control information
diffusion gets stronger, and the quicker a user spread
information, the easier it becomes for a user to stops
information from going viral. The greater a user’s between-
ness centrality is, the user can spread more quickly the
message to the entire network through the fewer users,
and the faster the user spreads information.

Copyright © Dr. Reem Essameldin 2023-2024