Graphs and Matrices

Unit -3
Based on – “Matthew Ganis, Avinash Kohirkar, Social Media
Analytics: Techniques and Insights for Extracting Business Value Out
of Social Media Pearson 2016”
Presenter Name – Shiv Tripathi
April 2025
Network Structures

Definition:
• Network structures define how individual entities (nodes)
are connected through relationships (edges) in a network.

• These structures help in analyzing the spread of

information, influence dynamics, and the formation of
communities. In social media analytics, network
structures enable the identification of key influencers,
understanding community behavior, and predicting
trends based on connectivity patterns.
Importance of Network Structures in Social Media Analytics:

• Influencer Identification: Understanding who drives

trends, engagement, and interactions in social networks.
• Community Detection: Recognizing groups of users
with shared interests and behaviors.
• Information Spread Analysis: Studying how news,
trends, or misinformation propagates.
• Fraud Detection: Identifying bot networks, fake
accounts, and spam clusters.
• Recommendation Systems: Personalizing content by
analyzing how users are interconnected.
Types of Network Structures:

1.Centralized Networks:
1.Have a dominant node that acts as the main point of interaction.
2.Examples: Facebook pages with millions of followers interacting
with one central brand account.
3.Pros: Efficient for information dissemination.
4.Cons: Vulnerable to single-point failure.
2.Decentralized Networks:
1.Composed of multiple hubs, ensuring no single node dominates.
2.Examples: Reddit communities, where multiple subreddits
function independently but are connected.
3.Pros: More resilient to failures and censorship.
4.Cons: Difficult to control or moderate content.
Types of Network Structures:

1.Distributed Networks:
1.Nodes have similar connectivity levels, making influence more
evenly spread.
2.Examples: Peer-to-peer networks like BitTorrent.
3.Pros: High resilience and fairness in content distribution.
4.Cons: Slower information spread compared to centralized networks.
2.Hierarchical Networks:
1.Structured in levels where information passes through layers before
reaching all nodes.
2.Examples: Corporate organizational charts, where communication
flows from top management to employees.
3.Pros: Clear roles and responsibilities.
4.Cons: Slower decision-making and communication bottlenecks.
Types of Network Structures:

Small-World Networks:
Characterized by short paths between nodes due to the
presence of influential hubs.
Examples: LinkedIn, where "six degrees of separation"
helps professionals connect globally.

Pros: Facilitates efficient networking.

Cons: May lead to echo chambers and reinforcement of
existing beliefs.
Visualization of Network Structures:

• Graph representations: Nodes as circles, edges as

lines connecting them.
• Adjacency matrices: Used for computational analysis of
relationships.
• Force-directed graphs: Visualizing clusters and
distances between entities.
Equivalence in Networks

Definition:
• Equivalence in networks refers to the similarity between
nodes based on their relationships and roles within the
network. It helps in categorizing users based on their
influence and connectivity patterns.
Types of Equivalence:

1.Structural Equivalence:
1.Two nodes have identical connections with the same set of
nodes.
2.Example: Employees reporting to the same manager have
similar structural equivalence in a corporate hierarchy.
3.Application: Used in job recommendation systems to find
similar professionals.
2.Automorphic Equivalence:
1.Nodes hold the same structural role but may not have identical
connections.
2.Example: CEOs of different companies play similar roles but are
not connected to the same employees.
3.Application: Identifying similar market leaders across industries.
Types of Equivalence:

Regular Equivalence:
• Nodes perform similar functions within different network
substructures.
• Example: Social media influencers in different niches
(fashion vs. tech influencers) share equivalent roles.
• Application: Understanding content creators’ impact in
different domains.
Equivalence in Social Media Analytics:

• Helps in influencer marketing by identifying users with

similar impact.
• Used in fraud detection by recognizing fake accounts
mimicking real ones.
• Aids in recommendation systems by identifying users
with common behaviors.
Homophily

Definition:
Homophily is the tendency of similar individuals to form
connections with each other. It plays a critical role in shaping
online communities and determining content reach.
Types of Homophily:
1.Status-Based Homophily:
1.Based on demographic attributes like age, gender, or education.
2.Example: LinkedIn users forming groups based on job roles.
2.Value-Based Homophily:
1.Based on shared interests, beliefs, or opinions.
2.Example: Twitter users following accounts that align with their
political views.
Impact of Homophily in Social Media Analytics:

• Creates echo chambers, limiting exposure to diverse

opinions.
• Influences advertising strategies, targeting users with
shared interests.
• Affects content virality, as like-minded users amplify
similar messages.
Clustering

Definition:
Clustering refers to the formation of tightly connected groups within a network.

Key Metrics:
• Clustering Coefficient: Measures the degree to which nodes cluster
together.
• Modularity: Determines the strength of division in a network.

Applications in Social Media Analytics:

• Identifying interest-based communities for targeted marketing.
• Detecting bot networks that cluster around misinformation campaigns.
• Understanding content engagement patterns for improving
recommendation systems.
Snowball Sampling

Definition:
A technique where existing users recruit new participants, growing
the sample exponentially.
Steps in Snowball Sampling:
1.Select initial subjects.
2.Ask them to refer more users.
3.Repeat the process to expand the dataset.
Applications:
• Influencer outreach programs.
• Identifying underground communities (e.g., dark web
networks).
• Analyzing niche social media trends.
Contact Tracing

Definition:
Tracking interactions between nodes to understand how information
or diseases spread.
Types:
• Forward Tracing: Tracking who an infected person has interacted
with.
• Backward Tracing: Identifying the original source of an infection
or rumor.
Applications:
• Misinformation tracking on social media.
• Understanding viral marketing campaigns.
• Epidemiology (e.g., COVID-19 tracking apps).
Random Walks in Networks

Definition:
A process where a path is randomly taken between nodes,
often used to analyze connectivity and structure.
Applications:
1.Search Engine Algorithms: Google’s PageRank for
ranking web pages.
2.Recommendation Systems: Netflix’s content
discovery algorithm.
3.Fraud Detection: Spotting abnormal transaction
behaviors.
Ego-Centered Networks

Definition: An ego-centered network focuses on one

individual (the "ego") and all the entities (called "alters")
that are directly connected to the ego. These networks
help understand the social environment and influence
around the ego.
Key Concepts:
• Ego: The central node in the network.
• Alters: All individuals directly connected to the ego.
• Ego Network Size: Number of alters.
• Density: How connected the alters are to each other.
Ego-Centered Networks

Applications:
• Social Media: Understanding a user's direct connections
(friends, followers).
• Marketing: Targeting individuals with high influence in
ego networks.
Example:
• In LinkedIn, your direct connections form your ego
network. The mutual connections among your contacts
determine network density.
• Diagram: Ego at the center with spokes connecting to
alters; some alters are interconnected.
Dominance Hierarchies

Definition: Dominance hierarchies represent ranked relationships

between individuals or groups, often seen in animal behavior but also
relevant in social and organizational networks.
Key Concepts:
• Hierarchy: A top-down structure where higher-ranked individuals
dominate lower-ranked ones.
• Social Roles: Roles defined based on dominance, authority, or influence.
Applications:
• Corporate structures: Organizational charts.
• Influencer marketing: Ranking users based on influence.
Example:
• In Twitter, celebrities or influencers have higher follower counts and exert
dominance in terms of content spread.
Third Party Records

Definition: These are external data sources that provide information about
interactions or affiliations, not collected directly from users but observed or
documented by others.
Types:
• Web analytics logs
• CRM data
• Government or academic databases
Applications:
• Data enrichment: Combine third-party records with user-generated data.
• Behavior analysis: Understanding patterns beyond social platforms.
Example:
• Using government data on education to correlate social media discussions about
literacy.
• Diagram: Two circles: one for social media data, another for third-party records,
with overlap representing combined insights.
Affiliation Networks

Definition: A type of bipartite network that connects

individuals to events, groups, or organizations they are
affiliated with.
Key Concepts:
• Bipartite Graph: Two types of nodes—users and
affiliations—with edges only between different types.
• Projection: Can be converted into one-mode networks
to analyze common affiliations.
Affiliation Networks

Applications:
• Event analysis: Identify user groups attending similar
events.
• Recommendation systems: Suggesting content based
on shared group memberships.
Example:
• Users in a Facebook group are connected via their
common membership.
• Diagram: Two columns of nodes: left (users), right
(groups/events), with lines connecting them.
Citation Networks

Definition: Citation networks represent scholarly articles or

patents as nodes, with directed edges from one item citing
another.
Key Concepts:
• Directed Edges: Show direction of citation (from citing to cited).
• Temporal Evolution: Older papers are cited by newer ones.
Applications:
• Research analytics: Discover influential research.
• Knowledge diffusion: Track how ideas spread over time.
Example:
• Google Scholar uses citation networks to rank and organize
publications.
Peer-to-Peer (P2P) Networks

Definition: P2P networks consist of nodes (peers) that share

resources without centralized coordination.
Key Concepts:
• Decentralization: No central authority controls the network.
• Symmetry: Each peer acts as both a client and server.
Applications:
• File sharing: BitTorrent networks.
• Blockchain: Distributed ledgers.
Example:
• Spotify initially used a hybrid P2P model to distribute music.