U2 IMP
Unit 2: Social Network Structure & Analysis
1. Basics of Social Network
-
Social Network: A structure showing how individuals (nodes) are connected (edges).
-
Example: Class of 25 students → each student = node; friendships = edges.
Key Components:
-
Nodes (Vertices): People, organizations, websites, etc. Properties include age, gender, interests, etc.
-
Edges (Links/Ties):
-
Undirected: Mutual (e.g., Facebook friends).
-
Directed: One-way (e.g., Twitter follow).
-
Weighted: Shows intensity (e.g., frequency of messages).
-
Ties:
| Aspect | Strong Ties | Weak Ties |
|---|---|---|
| Examples | Close friends | Acquaintances |
| Contact | Frequent | Occasional |
| Support | High | Low |
| Info Shared | Similar | New/diverse |
| Jobs | Limited | Broader scope |
2. Network Measures
Degree:
-
Number of connections per person.
-
Average Degree = Total connections / Total people.
Density:
-
Measures how connected the group is.
-
Formula: Density = Actual connections / Max possible connections.
-
Range: 0 (none) to 1 (fully connected).
Density Levels:
| Level | Range | Characteristics | Examples |
|---|---|---|---|
| Very Low | 0.0 - 0.2 | Sparse | Twitter networks |
| Medium | 0.4 - 0.6 | Moderate | Friend circles |
| Very High | 0.8 - 1.0 | Almost all connected | Wedding parties |
Centrality Measures:
| Measure | What it shows | Example |
|---|---|---|
| Degree | Most connected | Popular student |
| Betweenness | Bridge between groups | Knows nerds and athletes |
| Closeness | Reach everyone fast | Gossip center |
| Eigenvector | Knows influential people | Assistant to CEO |
3. Network Visualization
Layouts:
| Layout | Description | Best for |
|---|---|---|
| Spring | Nodes pull/push like springs | Revealing clusters |
| Circular | Arranged in a circle | Small groups |
| Hierarchical | Top-down | Company structure |
| Grid | Rows/columns | Easy comparison |
Big Network Issues:
-
Visualizing thousands of connections → becomes messy ("hairball").
-
Solution: Filter by community or interaction level.
4. Correlations in Networks
Triangles:
-
Three people all connected to each other.
-
Shows strong group cohesion.
Clustering Coefficient:
-
% of your friends who are also friends with each other.
-
Formula: Actual triangles / Possible triangles.
Assortativity:
-
Measures similarity in connections.
-
Range: -1 (opposites) to +1 (similar).
| Type | Meaning | Examples |
|---|---|---|
| Positive | Similar people connect | Same profession |
| Negative | Different people connect | Mentor-mentee |
5. Social Media Network Analytics
Terms:
-
Community: Group with more internal links.
-
Bridge: Connects different communities.
-
Hub: Highly connected node (e.g., influencer).
-
Reciprocity: Two-way connection (e.g., Facebook = 100%, Twitter = ~20%).
Platform Comparison:
| Platform | Type | Clustering | Purpose | Reciprocity |
|---|---|---|---|---|
| Undirected | High | Social | 100% | |
| Directed | Low | Info | 20–30% | |
| Undirected | Medium | Professional | 100% | |
| Directed | Medium | Content | 30–50% |
Tools for Analysis:
| Tool | Type | Use |
|---|---|---|
| Gephi | Visual platform | Easy graphs |
| NetworkX | Python lib | Coding-based |
| Cytoscape | Desktop app | Complex networks |
| R igraph | R library | Stats modeling |
| NodeXL | Excel Add-on | Easy for beginners |
| D3.js | JS library | Interactive visualizations |
Network Structures:
| Type | Description | Examples |
|---|---|---|
| Random | Random links | Early internet |
| Small-World | High clustering + shortcuts | Social networks |
| Scale-Free | Popular nodes get more links | Web, social media |
| Hierarchical | Clear levels | Military, corporate |