1.

Network Analysis
Network Analysis is the study of relationships and interactions among entities represented as
nodes (vertices) and links (edges). It helps understand how information, influence, or
resources flow in a network.
Networks can represent social connections, communication systems, biological interactions,
or transport routes.

Key Aspects:

 Nodes: Represent entities (people, devices, etc.)

 Edges: Represent relationships or interactions (friendship, communication, etc.)
 Directed/Undirected: Directional or mutual connections
 Weighted/Unweighted: Edges can have weights (e.g., distance, strength)

Applications:
Social media analysis, epidemiology, citation networks, and recommendation systems.

2. Graphs
Graphs are mathematical structures used to model pairwise relations between objects.
A graph G(V, E) consists of a set of vertices (V) and a set of edges (E) connecting them.

Types of Graphs:

 Directed Graph: Edges have direction (A → B).

 Undirected Graph: Edges have no direction (A — B).
 Weighted Graph: Each edge has a numerical value (e.g., distance).
 Bipartite Graph: Vertices divided into two groups where edges connect nodes from
different sets.

Example:
In a social graph, nodes represent users and edges represent friendships or follows.

Uses:
Pathfinding, shortest routes, influence propagation, clustering, etc.

3. Social Networks

A Social Network is a special type of graph that represents relationships among individuals
or groups.
Nodes are people (or accounts), and edges represent interactions like friendship,
communication, or collaboration.

Social network analysis processes social data structured in graphs. It involves the extraction of
several characteristics and graphics to describe the main properties of the network. Some
general properties of networks, such as the shape of the network degree distribution (defined
bellow) or the average path length, determine the type of network, such as a small -world
network or a scale-free network. A small-world network is a type of graph in which most nodes
are not neighbors of one another, but most nodes can be reached from every other node in a
small number of steps. This is the so-called small-world phenomenon which can be interpreted
by the fact that strangers are linked by a short chain of acquaintances. In a small-world network,
people usually form communities or small groups where everyone knows every- one else. Such
communities can be seen as complete graphs. In addition, most the community members have
a few relationships with people outside that community. However, some people are connected
to a large number of communities. These may be celebrities and such people are considered as
the hubs that are responsible for the small-world phenomenon. Many small-world networks are
also scale-free net- works. In a scale-free network the node degree distribution follows a power
law (a relationship function between two quantities x and y defined as y xn, where n is a
constant). The name scale-free comes from the fact that power laws have the same functional
form at all scales, i.e., their shape does not change on multiplication by a scale factor. Thus, by
definition, a scale-free network has many nodes with a very few connections and a small
number of nodes with many connections. This structure is typical of the World Wide Web and
other social networks. In the following sections, we illustrate this and other graph properties
that are useful in social network analysis.

Characteristics:

 Dense Clusters: Groups of tightly connected individuals.

 Bridges: Nodes connecting different communities.
 Hubs: Highly connected individuals (influencers).

Applications:

 Analyzing influencers on platforms like Twitter or Instagram.

 Studying information spread or rumor propagation.
 Detecting fake accounts or social bots.

4. Centrality
Centrality measures the importance or influence of a node within a network.
Different types of centrality highlight different aspects of importance.

The centrality of a node measures its relative importance within the graph. In this section we
focus on undirected graphs. Centrality concepts were first developed in social network analysis.
The first studies indicated that central nodes are probably more influential, have greater access
to information, and can communicate their opinions to others more efficiently [1]. Thus, the
applications of centrality concepts in a social network include identifying the most influential
people, the most informed people, or the most communicative people. In practice, what
centrality means will depend on the application and the meaning of the entities represented as
nodes in the data and the connections between those nodes. Various measures of the centrality
of a node have been proposed. We present four of the best-known measures: degree centrality,
betweenness centrality, closeness centrality, and eigenvector centrality.

Degree centrality is defined as the number of edges of the node. So the more ties a node has,
the more central the node is. To achieve a normalized degree centrality of a node, the measure
is divided by the total number of graph nodes (n) without counting this particular one (n 1).
The normalized measure provides proportions and allows us to compare it among graphs.
Degree centrality is related to the capacity of a node to capture any information that is floating
through the network. In social networks, connections are associated with positive aspects such
as knowledge or friendship.

Types of Centrality:

1. Degree Centrality: Number of direct connections a node has.

o High value → node is well-connected.
2. Closeness Centrality: Measures how close a node is to all others (based on shortest
paths).
o High value → node can reach others quickly.
3. Betweenness Centrality: Number of shortest paths passing through a node.
o High value → node acts as a bridge or broker.
4. Eigenvector Centrality: Measures influence based on connections to other
influential nodes.

Use: Identifying key influencers or control points in networks.

5. Drawing Centrality of Graphs

Drawing centrality involves visualizing graphs where the position or size of each node
reflects its centrality measure.

 High-degree nodes: Placed centrally or drawn larger.

 Low-centrality nodes: Placed at the periphery.
Visualization helps understand which nodes are influential and how information may
spread.

Tools Used:
Gephi, NetworkX (Python), Cytoscape, Pajek.

Example:
In a social network diagram, highly connected people appear in the center with larger circles.

[Link]
measures-in-a-graph-using-networkx-python/ ----- GO
THROUGH THIS LINK FOR PRACTICAL EXAMPLE
6. PageRank
PageRank is a link analysis algorithm developed by Google to rank web pages in search
results.
It measures the importance of each node in a directed graph based on incoming links.

A subvariant in the eigenvector centrality is PageRank. It’s defined as the measure of

directional influence of nodes and, thus, is most suited for directed graphs.
Since eigenvector centrality is suitable for undirected graphs, there wasn’t one for directed
graphs before PageRank entered the picture. It has applications in online platforms like
Twitter, which uses this centrality to offer recommendations of other accounts that a user can
follow.

PageRank is also used to detect flaws in the fraud detection system utilized in the insurance
and healthcare industries. The algorithm even predicts traffic workflows in public spaces and
streets by running over a graph that includes road intersections.

Concept:

 A page is important if many other important pages link to it.

 Each page distributes its importance equally among its outgoing links.

Formula (Simplified):
[
PR(A) = (1 - d) + d \sum_{i=1}^{n} \frac{PR(T_i)}{C(T_i)}
]
where

 (PR(A)) = PageRank of page A

 (T_i) = Pages linking to A
 (C(T_i)) = Number of outgoing links from (T_i)
 (d) = Damping factor (usually 0.85)

Applications:
Web ranking, citation analysis, identifying key nodes in social or biological networks.

7. Ego-Networks
An Ego-Network focuses on one specific node (the ego) and its immediate neighbors (the
alters).
It captures the local structure around an individual.

Ego-networks are subnetworks of neighbors that are centered on a certain node. In Facebook and LinkedIn,
these are described as “your network". Every person in an ego-network has her/his own ego-network and can
only access the nodes in it. All ego-networks interlock to form the whole social network. The ego-network
definition depends on the network distance considered. In the basic case, a distance of 1, a link means that
person A is a friends of person B, a distance of 2 means that a person, C, is a friend of a friend of A, and a
distance of 3 means that another person, D, is a friend of a friend of a friend of A. Knowing the size of an ego-
network is important when it comes to understanding the reach of the information that a person can transmit
or have access to.

Components:
  Ego: The focal node.
 Alters: Nodes directly connected to the ego.
 Ties among Alters: Connections between the neighbors themselves.

Uses:

 Studying individual influence or connectivity.

 Analyzing personal relationships in social networks.
 Understanding social support or isolation patterns.

Example:
In Facebook, your friends and how they are connected to each other form your ego-network.

8. Community Detection

Community detection identifies groups of nodes (clusters) that are more densely connected
to each other than to the rest of the network.
These communities reveal natural divisions within complex networks.

Common Algorithms:

 Modularity Optimization (Louvain Method): Maximizes the modularity score to

find clusters.
 Girvan–Newman Algorithm: Removes edges with high betweenness centrality to
separate communities.
 Spectral Clustering: Uses eigenvalues of adjacency matrices to detect structure.

Applications:

 Social media: Detecting friend groups or interest communities.

 Biology: Identifying functional gene or protein groups.
 Marketing: Finding customer segments or interest-based groups.

9. Example / Case Study

Case Study: Twitter Network Analysis

Objective:
To identify key influencers and communities in a Twitter discussion on climate change.

Steps:

1. Data Collection: Tweets, user mentions, retweets.

2. Graph Construction: Users as nodes, interactions as edges.
3. Centrality Analysis: Find influential users using degree and betweenness centrality.
4. PageRank: Identify authoritative accounts (e.g., news channels, scientists).
 5. Community Detection: Use Louvain algorithm to find topic clusters.
6. Visualization: Plot ego-networks of influencers to study sub-communities.

Outcome:
Revealed major influencer groups (activists, journalists, researchers) and their interaction
patterns.