Finding Influencers in a Social Network

In this advanced tutorial, you'll analyze a social network to find influential users, discover communities, and understand network dynamics. You'll apply multiple centrality metrics and learn when to use each one.

Level: Advanced Time: 25 minutes Prerequisites: Completed previous tutorials, understanding of graph concepts

What You'll Learn

• Applying multiple centrality metrics (degree, betweenness, PageRank, eigenvector)
• Finding mutual connections between users
• Analyzing verified vs non-verified users
• Understanding follow patterns and influence
• Extracting ego networks around key users

The Dataset

The social-network.gexf file represents a tech-focused social network:

• 23 users ranging from major influencers to new accounts
• 88 follow relationships (directed: follower → followed)
• Node attributes: display name, follower count, following count, verified status, category, join year
• Edge attributes: year the follow happened, interaction score

Step 1: Explore the Network

Let's understand what we're working with:

grph info examples/social-network.gexf

                   Graph Summary
┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Property          ┃ Value                        ┃
┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┩
│ File              │ examples/social-network.gexf │
│ Version           │ 1.3                          │
│ Mode              │ static                       │
│ Default Edge Type │ directed                     │
│ Node Count        │ 23                           │
│ Edge Count        │ 88                           │
└───────────────────┴──────────────────────────────┘

Node Attributes: category, display_name, followers, following, joined_year, verified

Key observation: this is a directed graph. A follow from A → B means A follows B (A sees B's content). The direction matters!

Step 2: Find Verified Accounts

Let's start by finding verified users - these are typically established influencers:

grph nodes examples/social-network.gexf --attr verified=True

You'll see accounts like @techguru, @codequeen, @airesearcher, and companies like @techcompany.

Compare with Non-Verified

grph nodes examples/social-network.gexf --attr verified=False

These include newer accounts like @juniordev, @bootcamper, @student, and @hobbyist.

Step 3: Find Top Influencers

Now the core question: who are the most influential users? Let's use multiple centrality metrics and compare.

Method 1: PageRank (Network Influence)

PageRank measures influence based on who follows you AND how influential those followers are:

grph centrality examples/social-network.gexf --type pagerank --top 5

   Pagerank Centrality (Top 5)
┏━━━━━━┳━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Rank ┃ Node        ┃ Score    ┃
┡━━━━━━╇━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ 1    │ codequeen   │ 0.125834 │
│ 2    │ techguru    │ 0.085864 │
│ 3    │ pythonista  │ 0.069127 │
│ 4    │ frontenddev │ 0.066747 │
│ 5    │ devtools    │ 0.064617 │
└──────┴─────────────┴──────────┘

Interesting! @codequeen ranks higher than @techguru despite having fewer raw followers. Why? Because she's followed by other influential accounts.

Method 2: In-Degree (Raw Follower Count in Network)

How many accounts in our network follow each user?

grph degree examples/social-network.gexf --direction in --top 5

This measures direct "popularity" within our network sample.

Method 3: Betweenness (Bridge Position)

Who connects different parts of the network?

grph centrality examples/social-network.gexf --type betweenness --top 5

Users with high betweenness bridge different communities. They might connect tech influencers with business accounts, or established users with newcomers.

Method 4: Eigenvector (Well-Connected to Well-Connected)

Similar to PageRank but considers the full network more equally:

grph centrality examples/social-network.gexf --type eigenvector --top 5

Compare all four methods - do the same accounts appear across metrics, or are there differences?

Step 4: Analyze User Categories

Let's see the different types of users:

Tech Accounts

grph nodes examples/social-network.gexf --attr category=tech

Business/Startup Accounts

grph nodes examples/social-network.gexf --attr category=business

Media Accounts

grph nodes examples/social-network.gexf --attr category=media

Company Accounts

grph nodes examples/social-network.gexf --attr category=company

Step 5: Find Mutual Connections

A key social network feature: finding users that both follow and are followed by someone (mutual follows), or finding common connections between two users.

Common Followers

Who do both @techguru and @codequeen follow? These are their common connections:

grph common-neighbors examples/social-network.gexf techguru codequeen

Common neighbors of techguru and codequeen

┃ ID             ┃ Label           ┃
├────────────────┼─────────────────┤
│ careerswitcher │ @careerswitcher │
│ podcasthost    │ @podcasthost    │

Both major influencers follow @podcasthost - probably a popular tech podcast!

Find More Common Connections

Try different pairs:

grph common-neighbors examples/social-network.gexf startupfounder vcpartner

What connections do the startup ecosystem accounts share?

Step 6: Analyze Follow Patterns

Let's look at who key accounts follow:

Who Does @techguru Follow?

grph neighbors examples/social-network.gexf techguru --direction out

This shows the accounts that @techguru follows (their outgoing edges).

Who Follows @codequeen?

grph neighbors examples/social-network.gexf codequeen --direction in

This shows accounts that follow @codequeen (incoming edges = followers).

Compare Follower Counts

Some accounts follow few but are followed by many (influencers), while others follow many but have few followers (consumers):

grph nodes examples/social-network.gexf --json | head -50

Look at the followers and following attributes - the ratio tells you a lot about an account.

Step 7: Trace Connection Paths

How are two users connected? Let's trace paths:

New User to Influencer

How is @student connected to @techguru?

grph path examples/social-network.gexf student techguru

This shows the follow chain from @student to @techguru.

Can Newcomers Reach Everyone?

grph has-path examples/social-network.gexf student techcompany

Check if there's any path from a new account to a major company account.

Step 8: Extract Ego Networks

An "ego network" is the network around a specific user. Let's extract @codequeen's network:

grph ego examples/social-network.gexf codequeen --radius 1

This shows @codequeen and everyone directly connected (followers and following).

Expand the Network

Include connections of connections:

grph ego examples/social-network.gexf codequeen --radius 2

This reveals the extended network - often called "friends of friends."

Step 9: Find Newcomer Growth Paths

Let's help new accounts understand who to follow for growth.

Who Should @bootcamper Follow?

Find influential accounts that might follow back (those who follow newcomers):

# Who does @codequeen follow? (She follows some newcomers)
grph neighbors examples/social-network.gexf codequeen --direction out

Notice that @codequeen follows @juniordev and @careerswitcher - she engages with newcomers!

Trace Successful Newcomers

Who does @careerswitcher follow? They grew from 0 to 8500 followers:

grph neighbors examples/social-network.gexf careerswitcher --direction out

Step 10: Network Statistics

Finally, let's understand the overall network structure:

grph stats examples/social-network.gexf

Key metrics to look for:

• Density: How interconnected is the network?
• Average path length: How many "degrees of separation" on average?
• Clustering: Do friends tend to be friends with each other?

Comparing Centrality Metrics

Here's a summary of when to use each metric:

Metric	Measures	Best For
PageRank	Influence from quality followers	Finding true influencers
In-Degree	Raw follower count	Simple popularity
Out-Degree	How many they follow	Finding active engagers
Betweenness	Bridge position	Finding connectors
Eigenvector	Connected to connected	Finding inner circle

What You've Learned

You've mastered social network analysis:

• Multiple centrality metrics: Understanding different types of influence
• Mutual connections: Finding shared relationships
• Follow pattern analysis: Understanding account behavior
• Ego networks: Extracting networks around key users
• Path tracing: Understanding connection chains

Real-World Applications

These techniques apply to:

• Marketing: Finding influencers for campaigns
• Community building: Understanding network structure
• HR/Recruiting: Mapping professional networks
• Research: Academic citation and collaboration networks
• Security: Detecting bot networks or coordinated behavior

Next Steps

• Read the Centrality Metrics Explained for deeper understanding
• Try the How-to: Find Influencers guide for a quick reference
• See how to Export for Visualization to create network diagrams

Exercises

1. Find the account with the highest follower-to-following ratio: Use grph nodes --json and calculate the ratio.

2. Who are the earliest adopters in the network?

   grph nodes examples/social-network.gexf --json | grep "joined_year" | sort

3. Find accounts that follow each other (mutual follows): For two accounts A and B, check if both A → B and B → A exist using grph has-path.

4. Create a visualization of @techguru's ego network:

   grph ego examples/social-network.gexf techguru --radius 1 --output techguru-network.gexf
   grph export techguru-network.gexf --format json --output techguru.json