Introduction to Social Network Analysis

Social Network Analysis (SNA)

Refers to the quantitative and qualitative analysis of social network data.

It describes networked structures in terms of nodes (individual actors, users, or things within the network) and the links, edges, or ties (relationships or interactions) connecting them.

SNA steps

1. Problem identification

2. Selection of data sources

3. Data cleaning

4. Data transformation

5. Mining (analysis)

6. Analysis of results

network vs graph

network → to refer to the structure of social relationships.

graph → the mathematical representation of a network.

Degree of a node

The number of connections that node has.

A high-degree node often has more influence because it is connected to many others.

undirected graph

Relationship goes both ways.

Example: Alice ——— Bob (no arrows)

If Alice is friends with Bob,

Bob is automatically friends with Alice.

directed graph

Relationship goes one way.

Example: Alice → Bob (Now the direction matters)

Alice follows Bob.

Bob does not have to follow Alice.

Symmetric/not symmetric Adjacency Matrix

Undirected graph

If Alice is connected to Bob, Bob must also be connected to Alice.

Therefore the adjacency matrix is symmetric.

If (Alice, Bob) = 1, then automatically (Bob, Alice) = 1

Directed graph

Not necessarily. Alice may follow Bob, while Bob doesn't follow Alice.

Therefore (Alice, Bob) = 1, (Bob, Alice) = 0

The matrix is not symmetric.

weighted graphs

Sometimes connections are stronger than others.

Instead of only writing 0 = no connection; 1 = connection, we can assign a weight.

Example:

Emails exchanged

Alice ↔ Bob → 200 emails

Alice ↔ Claire → 5 emails

Alice —- 200 —- Bob

Alice —- 5 —- Claire

self-organisation

Large patterns emerge without anyone planning them.

Example irl: psychology students become friends with each other without anyone telling them “become friends with each other.” Nobody organised them.

small-world property

Almost everyone can be reached through only a few connections.

preferential attachment

Popularity attracts even more popularity.

scale-free network

A network where most nodes have few connections and a few nodes (hubs) have many.

computing network measures

Researchers calculate numbers that describe the overall structure of the network.

Imagine everyone in a network. For every pair of people, find the shortest route between them. Average all those routes. If the average is small, information can spread quickly.

community detection

Finds groups of nodes that are strongly connected to each other.

e.g., friend groups, university network.

link prediction

Which new connections are likely to appear?

information propagation

Information spreads through networks much like diseases.

Researchers study:

• Who starts the spread

• How quickly it spreads

• Which communities receive it

• When it stops

social influence maximisation

SNA tries to identify the smallest group of people that will influence the largest number of others.

This is useful for:

• marketing

• public health campaigns

• election campaigns

• promoting healthy behaviours

at the end of the chapter, you can analyse:

• Network structure (who connects to whom)

• Content (what they actually post)