ANSWER KEY

Social Network Analysis – Question Bank

[Link] – Computer Science / AI & ML Specialization
TKREC – R22 Regulation
Mapped from: Analyzing Social Media Networks with NodeXL

SNA Answer Key | Page 1

SNA Answer Key | Page 2
UNIT I – Introduction to Social Media and Social
Network Analysis

Part A: 1-Mark Answers

Q1. Define 'Social Media' as per the framework discussed in Social Network Analysis.
Answer:
Social media refers to a set of computer-network based tools that support social interaction
between users. It transforms broadcast (one-to-many) into dialog (many-to-many),
encompassing tools like email, blogs, microblogs, wikis, social networking sites, and video-
sharing platforms.
Q2. What does the acronym SNA stand for?
Answer:
SNA stands for Social Network Analysis — a set of methods and techniques for measuring,
mapping, and modeling collections of connections among individuals, groups, or entities.
Q3. What is a 'node' (vertex) in a social network graph?
Answer:
A node (also called a vertex, agent, or entity) is the fundamental unit in a network graph. It
can represent people, organizations, web pages, keywords, or events. In social media
networks, nodes typically represent user accounts.
Q4. What is an 'edge' (tie) in a social network?
Answer:
An edge (also called a tie, link, connection, or relationship) connects two vertices in a
network. It represents a relationship such as friendship, email exchange, a retweet, or a
hyperlink. Edges can be directed or undirected, and weighted or unweighted.
Q5. Distinguish between synchronous and asynchronous social media
communication.
Answer:
Synchronous communication requires both parties to interact at the same time (e.g., live chat,
video conferencing, multiplayer games). Asynchronous communication allows participation at
different times (e.g., email, discussion forums, voicemail). Social media increasingly blurs this
distinction.
Q6. What is a 'sociogram'?
Answer:
A sociogram (also called a network graph) is a visual map of social relationships where
individuals are represented as nodes and connections between them as lines or arrows. It
was pioneered by Jacob Moreno in the 1930s to study group relationships.
Q7. Name any two alternative terms used for 'social media' systems.
Answer:
Two alternative terms include: (1) 'Web 2.0' – referring to read/write web tools that enable
user participation, and (2) 'Computer-Supported Cooperative Work (CSCW)' – systems
designed to support group collaboration over networks.
Q8. What is the difference between a directed and undirected edge?
Answer:

SNA Answer Key | Page 3

 A directed edge (asymmetric edge) has a clear origin and destination — e.g., one user follows
another on Twitter (one-way). An undirected edge (symmetric/mutual edge) simply exists
between two entities with no directional distinction — e.g., Facebook friendship, which
requires mutual agreement.
Q9. Define 'degree centrality' in one sentence.
Answer:
Degree centrality is a simple count of the total number of direct connections (edges) linked to
a vertex, serving as a measure of a node's popularity or activity within the network.
Q10. Who is credited as the pioneer who developed the concept of sociometry?
Answer:
Jacob Moreno is credited as the pioneer of sociometry and the sociogram. In 1934, he
published 'Who Shall Survive?' which catalyzed modern social network analysis by
introducing visual maps of group relationships and formal network metrics.

Part B: 10-Mark Answers

Q1. Explain the evolution and definition of Social Media. How does it differ from
traditional media? Discuss the six key design dimensions used to characterize social
media systems with suitable examples.
Answer:
Definition & Contrast with Traditional Media:
Social media refers to computer-network based tools that support social interaction between
users. It transforms broadcast (one-to-many) into dialog (many-to-many). Traditional media
(TV, newspapers, books) delivers content to mass audiences without enabling user creation
or sharing. Social media is interactive, participatory, and user-driven.
Six Key Design Dimensions (Hansen, Shneiderman & Smith):
• Size of Producer and Consumer Population: Who creates and who consumes? e.g., Twitter
allows few to broadcast to millions; email is typically one-to-one.
• Pace of Interaction: Synchronous (chat, video conferencing) vs. asynchronous (email,
forums). Twitter has blurred this with near-real-time replies.
• Genre of Basic Elements: The type and format of digital objects — tweets (280 characters),
wiki pages (structured text), YouTube videos, Instagram photos.
• Control of Basic Elements: Who can create, edit, read, or moderate? Open systems
(Wikipedia) vs. closed systems (private forums). Barriers to entry affect community
governance.
• Types of Connections: How users are linked — friendship (Facebook), following (Twitter),
co-editing (Wikipedia), hyperlinking (WWW).
• Retention of Content: How long content is visible — ephemeral (Snapchat) vs. permanent
archives (email lists, Wikipedia).
These six dimensions provide a systematic framework to compare social media tools and
understand the unique networks they generate.
Q2. What is Social Network Analysis (SNA)? Explain the 'network perspective' and how
it differs from traditional attribute-based analysis. Discuss real-world applications of
SNA across various domains.
Answer:
Definition of SNA:
Social Network Analysis (SNA) is the study of social structures using network theory —
measuring, mapping, and modeling collections of connections (ties) among individuals,
groups, or entities. It focuses on relationships rather than the attributes of individuals.

SNA Answer Key | Page 4

 Network Perspective vs. Attribute-Based Analysis:
Traditional analysis focuses on individual attributes (age, income, education). SNA focuses on
the structure of relationships — who is connected to whom and how. Network analysis argues
that outcomes (getting a job, adopting innovations, catching diseases) are often better
explained by network position than personal characteristics. As the textbook states: 'Know
who is often more important than know how.'
Real-World Applications:
• Sociology: Mapping community structures, identifying social roles.
• Business: Identifying influencers, organizational communication analysis, viral marketing.
• Public Health: Modeling epidemic spread (e.g., COVID-19 contact networks).
• Law Enforcement: Mapping criminal networks via email trails.
• Engineering: Designing robust power grids, computer networks.
• Political Science: Analyzing voting patterns, political polarization on Twitter.
• Computer Science: Google's PageRank algorithm for ranking web pages.
Q3. Describe the primary building blocks of a social network: vertices and edges.
Explain the different types of edges with examples from social media platforms. (Draw
and explain diagram)
Answer:
Vertices (Nodes):
Vertices are the entities in a network. They represent people (Twitter users, Facebook
friends), content (web pages, wiki articles), or locations. Attribute data (e.g., number of
followers) can be overlaid onto vertices using size, color, or shape in visualizations.
Edges (Ties):
Edges are the connective tissue of networks — representing relationships between vertices.
Types of Edges:
• Directed Edge (Asymmetric): Has a clear origin and destination. Example: User A follows
User B on Twitter — an arrow points from A to B.
• Undirected Edge (Symmetric): Mutual relationship with no directionality. Example: Facebook
friendship — both users must agree.
• Weighted Edge: Carries a numerical value indicating strength or frequency. Example:
Number of emails exchanged between two users; thicker lines represent stronger ties.
• Unweighted (Binary) Edge: Simply indicates presence or absence of a tie. Example: Either a
Facebook friendship exists (1) or it does not (0).
Diagram:
(Draw and explain diagram) — Draw three nodes A, B, C. A→B (directed), A↔C (undirected
mutual), A→B with thickness 3 (weighted). Label each edge type clearly with arrows.
Q4. Explain the three types of social network data representations: adjacency matrix,
edge list, and network graph. Illustrate with a small example network. (Draw and
explain diagram)
Answer:
(1) Adjacency Matrix:
A grid where rows and columns represent all vertices. A '1' in cell (i,j) means vertex i has a
connection to vertex j; '0' means no connection. For the example network {Ann→Bob,
Ann→Carol, Carol→Ann}: Matrix shows Ann row has 1s in Bob and Carol columns; Carol row
has 1 in Ann column; Bob row is all 0s. Limitation: becomes very large and sparse for big
networks.
(2) Edge List:
A two-column table listing every connection: Vertex1 | Vertex2. Same network: Ann|Bob,
Ann|Carol, Carol|Ann. More efficient than the matrix for sparse networks since only existing
connections are stored.
(3) Network Graph (Sociogram):

SNA Answer Key | Page 5

 Visual representation where nodes are drawn as circles/icons and edges as lines/arrows.
Immediately reveals structural patterns (clusters, bridges, hubs) that are hidden in tabular
data. NodeXL uses this for visualization.
(Draw and explain diagram) — Draw the 3-node directed graph with Ann, Bob, Carol. Show
arrows as described above, with labels.
Q5. Discuss the types of social networks: egocentric, partial, and full networks. What
are unimodal, multimodal, and affiliation networks? Provide examples.
Answer:
By Scope:
• Egocentric Network: Centers on a single individual (ego) and their direct connections
(alters). The 1-degree ego network includes ego + alters; the 1.5-degree also includes
connections among alters (e.g., your Facebook friend circle).
• Partial Network: A selected slice of a larger network. Example: Twitter network of the 100
U.S. Senate members — a topic-centric subset.
• Full/Complete Network: Contains all entities matching a criterion and all connections
between them. Example: The complete Twitter platform network (all users and their
connections).
By Vertex Type:
• Unimodal Network: Contains only one type of vertex — e.g., user-to-user Twitter mention
network.
• Multimodal Network: Contains multiple types of vertices — e.g., a network connecting
Marvel movie characters to the movies they appear in.
• Bimodal Affiliation Network: Exactly two types of vertices — e.g., Wikipedia editors and the
articles they edit. No direct editor-to-editor edges; they are linked indirectly through shared
article edits. Can be transformed into a unimodal editor-to-editor network.
Q6. Trace the historical development of Social Network Analysis from Moreno's
sociometry to modern computational approaches.
Answer:
1930s – Foundations (Jacob Moreno):
Jacob Moreno pioneered sociometry and the sociogram. His 1934 work 'Who Shall Survive?'
introduced visual maps of social relationships among schoolchildren and military recruits.
Harvard's group (Lloyd Warner, Elton Mayo) studied factory worker relationships at Western
Electric's Hawthorne Plant.
1950s–1970s – Mathematical Formalization:
Paul Lazarsfeld and colleagues added quantitative metrics. Mark Granovetter (1973)
discovered the 'Strength of Weak Ties' — that casual acquaintances ('weak ties') provide
more novel information (like job leads) than close friends. Stanley Milgram's (1967) 'Six
Degrees of Separation' experiment showed the small-world property of social networks.
1970s–1990s – Expansion:
Barry Wellman studied 'networked individualism.' INSNA (International Network for Social
Network Analysis) founded in 1977. Kleinberg's HITS algorithm (1990s) and Brin & Page's
PageRank (Google) applied network analysis to the WWW.
2000s–Present – Social Media Era:
The explosion of social media (Facebook, Twitter, YouTube) generated vast, machine-
readable network datasets. Tools like NodeXL democratized SNA for non-programmers.
Watts & Strogatz modeled 'small world' networks mathematically; Barabasi & Albert
developed 'scale-free' network models. Network science now spans all academic disciplines.
Q7. What is a multiplex network? How does it differ from a uniplex network? Describe
structural holes and brokerage.
Answer:

SNA Answer Key | Page 6

 Uniplex vs. Multiplex Networks:
A uniplex network contains only one type of connection between vertices (e.g., only 'follow'
relationships on Twitter). A multiplex network contains multiple types of connections between
the same pair of vertices — e.g., the Twitter network shown in Figure 3.1 includes both 'reply-
to' and 'mention' edges. Multiplex data is often condensed into a uniplex network for analysis
by combining edge types.
Structural Holes:
A structural hole is a gap in the network — a missing connection between two otherwise
separate groups or individuals. Ron Burt demonstrated that individuals who bridge structural
holes (i.e., who connect otherwise disconnected groups) are promoted faster and gain
competitive advantages because they control the flow of information between groups.
Brokerage:
Brokerage refers to the role of individuals who bridge across structural holes, acting as
intermediaries between groups. Betweenness centrality is the primary metric used to identify
brokers — it measures how often a vertex lies on the shortest path between two other
vertices. Brokers in organizations can translate across silos, identify synergies, and are
strategically valuable.

UNIT II – NodeXL: Layout, Visual Design, Network

Metrics, and Clustering

Part A: 1-Mark Answers

Q1. What does NodeXL stand for, and under which software platform does it operate?
Answer:
NodeXL stands for Network Overview, Discovery and Exploration for Excel. It is a free, open-
source add-on template that operates within Microsoft Excel (versions 2007 and later). It is
maintained by the Social Media Research Foundation (SMRF).
Q2. Name the two main interface sections in NodeXL.
Answer:
NodeXL consists of: (1) The Spreadsheet — where network data (edges, vertices, groups) is
stored in multiple worksheets; and (2) The Graph Pane (Document Actions) — the visual
panel that displays the network graph visualization.
Q3. What is the Fruchterman-Reingold algorithm used for in NodeXL?
Answer:
The Fruchterman-Reingold algorithm is a force-directed graph layout algorithm used to
automatically arrange vertices in the graph pane. It simulates physical forces (attraction and
repulsion) so that connected nodes are drawn closer together and unconnected nodes are
pushed apart, creating more readable and aesthetically organized network visualizations.
Q4. Define 'network density' as a graph-level metric.
Answer:
Network density is the ratio of the number of existing edges to the total number of possible
edges in a network. Formula: Density = (Actual edges) / (Maximum possible edges). It

SNA Answer Key | Page 7

 captures concepts like cohesion and solidarity — a density of 1.0 means every vertex is
connected to every other vertex.
Q5. What is 'betweenness centrality'?
Answer:
Betweenness centrality measures how often a given vertex lies on the shortest path
(geodesic path) between two other vertices. It is a 'bridge score' — high betweenness
indicates a vertex acts as an important broker or gatekeeper connecting different parts of the
network.
Q6. What does 'clustering coefficient' measure for a vertex?
Answer:
The clustering coefficient measures the density of the 1.5-degree egocentric network of a
vertex — i.e., how interconnected a vertex's immediate neighbors are. If all of a vertex's
friends know each other, its clustering coefficient is high (close to 1). If none of its friends
know each other, it is low (close to 0).
Q7. What is an 'isolate' in a network graph?
Answer:
An isolate is a vertex that has no connections to any other vertex in the network (degree = 0).
In NodeXL, isolates are visually separated from the rest of the network and can be laid out in
separate boxes to avoid cluttering the main graph.
Q8. Define 'closeness centrality' in one sentence.
Answer:
Closeness centrality measures the average distance (number of hops) from a given vertex to
every other vertex in the network — vertices with low closeness centrality are 'closer' to all
others, while high values indicate peripheral positions.
Q9. What is the PageRank algorithm and which company popularized it?
Answer:
PageRank is a variant of Eigenvector centrality for directed networks developed by Sergey
Brin and Larry Page, who founded Google. It ranks web pages based on (1) the number of
inbound links, (2) the quality (PageRank) of those linkers, and (3) the link propensity of the
linkers. Pages cited by many high-quality pages receive higher PageRank scores.
Q10. What is a 'connected component' in a network?
Answer:
A connected component is a subset of vertices in a network where every vertex is reachable
from every other vertex within that subset through some path of edges, but the subset has no
connections to vertices outside it. A network may consist of multiple disconnected
components.

Part B: 10-Mark Answers

Q1. Describe the NodeXL interface in detail. Explain how to import edge data, display a
network graph, and highlight specific vertices or edges.
Answer:
NodeXL Overview:
NodeXL is a Microsoft Excel add-on template maintained by the Social Media Research
Foundation. It comes in two versions: NodeXL Basic (free) and NodeXL Pro (subscription,
adds social media importers, advanced metrics, content analysis).
Interface Components:

SNA Answer Key | Page 8

 • Spreadsheet: Multiple worksheets — Edges (each row = one connection), Vertices (each
row = one entity), Groups (cluster assignments), Group Vertices, Overall Metrics.
• Graph Pane (Document Actions): Displays the visual network graph. Can be resized by
dragging or undocked for multi-monitor use.
• NodeXL Ribbon: Organized into sections — Data, Graph, Visual Properties, Analysis,
Options, Show/Hide, Help.
Importing Data:
• Manual Entry: Type edge pairs (Vertex1, Vertex2) directly into the Edges worksheet.
• Import Menu: Use Import drop-down to load data from UCINET, Pajek, GraphML files, or
directly from social media APIs (Twitter, YouTube) via NodeXL Pro importers.
• From Open Workbook: Import an edge list or adjacency matrix from another open Excel file.
Displaying the Graph:
Click 'Show Graph' button above the graph pane. NodeXL reads the Edges worksheet and
auto-populates the Vertices worksheet with unique vertices. The button changes to 'Refresh
Graph' after first use.
Highlighting:
The spreadsheet and graph pane are coupled — clicking a row in the Edges worksheet
highlights the corresponding edge in the graph (shown in red). Clicking a vertex row highlights
the vertex. Multiple rows can be selected with Ctrl+Click.
Q2. Explain the various network layout algorithms available in NodeXL. What are the
best practices for creating readable network visualizations? (Draw and explain
diagram)
Answer:
Layout Algorithms:
• Manual Layout: Drag vertices to custom positions. Useful for small networks or when
emphasizing specific relationships.
• Fruchterman-Reingold: Force-directed algorithm; simulates attraction (connected nodes)
and repulsion (all nodes). Produces organic, clustered layouts. Settings can be adjusted
(iterations, C value).
• Harel-Koren Fast Multiscale: Efficient for larger networks; produces readable layouts quickly
by operating at multiple scales.
• Grid, Circle, Spiral, Sine Wave: Simple geometric layouts for specific presentation needs.
Readability Heuristics (Netviz Nirvana – Shneiderman):
• Make every vertex visible (avoid occlusion/overlap).
• Every vertex's degree should be countable.
• Every edge can be traced from source to destination.
• Clusters and outliers are visually identifiable.
• Separate isolates into boxes at the bottom.
• Lay out each cluster in its own box for clarity.
(Draw and explain diagram) — Show a before/after layout comparison: tangled hairball vs.
organized Fruchterman-Reingold layout with visible clusters.
Q3. Describe the process of visual design in NodeXL — how can you customize vertex
color, size, shape, opacity, and labels to convey attribute data?
Answer:
Visual Properties in NodeXL:
NodeXL allows network attribute data to be mapped to visual properties of vertices and
edges. This transforms abstract numbers into intuitive visual cues.
Vertex Visual Properties:
• Size: Map to a metric (e.g., degree centrality or follower count). Larger vertex = more
connections. In the US Senate Twitter example, vertex size maps to follower count.

SNA Answer Key | Page 9

 • Color: Encode group membership or attribute categories (e.g., political party — blue for
Democrats, red for Republicans).
• Shape: Different shapes (circle, square, diamond) can indicate different vertex types or
roles.
• Opacity: Fade less important vertices; emphasize key nodes.
• Label: Display user names, IDs, or metrics. Labels can be positioned around vertices.
• Image: Use actual profile photos (as in the Senate Twitter network visualization).
Edge Visual Properties:
• Thickness (Width): Maps to edge weight — thicker edges indicate more frequent
interactions.
• Color: Distinguishes edge types (e.g., retweet vs. mention).
• Opacity: Reduces visual clutter for dense networks.
Q4. Explain the following aggregate network metrics with formulas and examples:
Vertex count, Edge count, Density, Number of connected components, and Graph
diameter.
Answer:
• Vertex Count: Total number of entities (nodes) in the network. Example: 100 senators = 100
vertices.
• Edge Count: Total number of connections in the network. Example: 450 mention
relationships among senators.
• Density: Density = E / (V × (V−1)) for directed networks, where E = edges, V = vertices.
Example: 450/(100×99) ≈ 0.045 means only 4.5% of all possible connections exist — a
sparse network typical of real social networks.
• Number of Connected Components: Count of disconnected sub-networks within the graph.
Example: A Twitter hashtag network may have one large main component and several
isolated pairs or singletons.
• Graph Diameter: The longest of all shortest paths between any pair of vertices in the
network — the 'worst case' distance across the network. Example: If the Senate network
diameter = 3, any senator is at most 3 hops from any other.
Q5. Explain the following vertex-level centrality metrics with examples: Degree
centrality, Betweenness centrality, Closeness centrality, and Eigenvector centrality.
Answer:
• Degree Centrality: Count of direct connections. For directed networks: In-degree
(connections pointing IN — e.g., being mentioned) and Out-degree (connections pointing
OUT — e.g., number of mentions sent). A senator with 50 replies received has an in-
degree of 50.
• Betweenness Centrality: Fraction of shortest paths between all vertex pairs that pass
through the vertex. High betweenness = bridge/broker between communities. A journalist
who connects politicians from both parties would have high betweenness. Formula: BC(v)
= Σ(σst(v)/σst) for all pairs s≠v≠t.
• Closeness Centrality: Inverse of the average shortest path distance from the vertex to all
others. Low closeness score = vertex is close to many others (counterintuitively, lower
average distance = more 'central'). A moderator in a discussion forum who interacts with
all subgroups has low closeness scores.
• Eigenvector Centrality / PageRank: A vertex is important if it is connected to other important
vertices. Few connections to highly-connected hubs can outweigh many connections to
peripheral nodes. Google's PageRank uses this for web pages: a page linked by CNN has
higher PageRank than a page linked by obscure blogs.
Q6. What is clustering and grouping in NodeXL? Explain community detection
algorithms. How can visual separation of clusters aid in network interpretation? (Draw
and explain diagram)
Answer:

SNA Answer Key | Page 10

 Clustering & Grouping in NodeXL:
Clustering identifies subsets of vertices that are more densely connected to each other than
to the rest of the network. NodeXL automates this using community detection algorithms.
Groups are stored in the Groups worksheet and visually separated in the graph pane.
Community Detection Algorithms:
• Clauset-Newman-Moore: Optimizes modularity to find dense subgraphs. Fast and effective
for large networks.
• Wakita-Tsurumi: Similar modularity-based approach, efficient for medium-large networks.
• Girvan-Newman: Removes edges with highest betweenness iteratively to reveal community
structure.
• Connected Components: Groups vertices that are reachable from one another — a basic
form of clustering.
Visual Benefits:
• Separate boxes for each cluster: Clarify boundaries between communities.
• Color-coding groups: Instantly distinguish clusters.
• Show intra-cluster edges vs. inter-cluster edges separately: Reveals bridge relationships.
(Draw and explain diagram) — Draw a network with two clear clusters (A-B-C densely
connected; D-E-F densely connected) with one weak bridge edge between clusters, color
each group differently.
Q7. Describe the process of preparing, filtering, and cleaning network data in NodeXL
for large, complex networks.
Answer:
Data Preparation:
• Import via NodeXL importers (Twitter, email, YouTube APIs) or manual entry.
• Resolve entity duplication: Multiple email addresses for the same person must be merged
using Find & Replace or a Lookup Table approach.
Filtering Strategies:
• Time-based filtering: Include only messages/interactions within a specific date range.
• Sender/Receiver filtering: Focus on specific individuals or departments.
• Keyword filtering: Include only edges involving certain topics (e.g., only emails containing
'cybersecurity').
• Degree-based filtering: Remove low-degree vertices (e.g., keep only vertices with degree >
2) to reduce noise.
Cleaning:
• Remove self-loops (edges where a vertex points to itself).
• Delete duplicate edges or aggregate them into weighted edges.
• Normalize edge weights for consistent visualization.
Managing Large Networks:
Very large networks (100,000+ edges) may require pre-filtering before visualization. Use
NodeXL's 'Get Vertices from Edges' without displaying the graph first. Consider sampling
strategies or focusing on egocentric subnetworks of key vertices.

UNIT III – Case Studies: Email Networks and Thread

Networks (Twitter)

SNA Answer Key | Page 11

Part A: 1-Mark Answers

Q1. Why is email described as 'the lifeblood of modern communication'?

Answer:
Email is described as the lifeblood of modern communication because it integrates into
everyday professional and personal life more deeply than any other tool. It supports flexible
(plain text to rich formatting), asynchronous, broadcast, threaded communication and became
the 'killer app' of the early internet (ARPANET). By 1973, three-fourths of all ARPANET traffic
was email.
Q2. What type of network is created when email is sent from one person to another?
Answer:
A directed, weighted network is created. Each email address/person is a vertex. A directed
edge points from the sender to the receiver. The edge is weighted by the number of
messages exchanged — more messages = higher edge weight (represented by thicker edges
in NodeXL visualizations).
Q3. Define a 'thread network' in the context of email and message boards.
Answer:
A thread network is a network created by the reply structure of messages in email threads or
online message boards. Edges connect message authors based on reply-to relationships —
when Author A replies to Author B's message, an edge is formed from A to B. Threads reveal
conversation patterns, active participants, and discussion starters.
Q4. What does the Twitter API stand for?
Answer:
Twitter API stands for Twitter Application Programming Interface. It is a set of protocols and
tools that allows developers and researchers to programmatically access Twitter data —
retrieving tweets, user profiles, follower lists, and network connections. NodeXL Pro uses the
Twitter API to import network data directly.
Q5. What is a 'tweet' and what is its maximum character length?
Answer:
A tweet is a short public message posted on Twitter (now X). The original maximum character
limit was 140 characters, later expanded to 280 characters. The short format promotes rapid
information sharing, hashtag use, and frequent interactions that create rich network
structures.
Q6. Distinguish between 'reply-to' and 'mention' relationships on Twitter.
Answer:
A reply-to relationship occurs when a user directly responds to another user's tweet using the
Reply function, creating a directed edge from responder to the original poster. A mention
relationship occurs when a user includes another user's @handle anywhere in their tweet, not
necessarily as a direct reply. Both create directed edges and are used as different edge types
in Twitter network analysis.
Q7. What is a 'retweet' and what network relationship does it imply?
Answer:
A retweet (RT) is when a user shares another user's tweet with their own followers. It implies
a directed edge from the retweeter to the original author. Retweets are key indicators of
information diffusion and content endorsement — high retweet count indicates viral content or
influential authors.
Q8. Define 'hashtag' and its role in Twitter network formation.
Answer:

SNA Answer Key | Page 12

 A hashtag is a word or phrase preceded by the '#' symbol used to categorize and search
tweets by topic (e.g., #MachineLearning). Hashtags create topic-centric networks — all users
tweeting with the same hashtag are part of a shared conversation. NodeXL can collect
networks by searching a specific hashtag, connecting users who tweet about the same topic.
Q9. What is a 'mailing list' in the context of thread network mapping?
Answer:
A mailing list (or Listserv) is a community email collection where messages sent to a list
address are automatically forwarded to all subscribed members. Unlike personal email,
mailing list archives are often public or semi-public. Thread networks from mailing lists reveal
community roles (experts, moderators, lurkers) and discussion dynamics over time.
Q10. Name one tool used to import and analyze Twitter network data in NodeXL.
Answer:
NodeXL Pro's built-in Twitter Network Importer (powered by the Twitter API) allows direct
import of Twitter conversations, hashtag networks, user-to-user mention/reply networks, and
follower networks. The Social Media Research Foundation's NodeXL Graph Gallery also
provides pre-built Twitter network datasets.

Part B: 10-Mark Answers

Q1. Explain how email networks are constructed and analyzed using NodeXL. What
types of connections are formed in email networks, and what do they reveal about
organizational communication?
Answer:
Email Network Construction:
In an email network, vertices represent email addresses (and corresponding people).
Directed, weighted edges are created when a message is sent from one address to another
— an arrow points from sender to receiver, and edge weight reflects the number of messages
exchanged.
NodeXL Import Process:
NodeXL's 'Import from Email Network' feature uses Windows built-in indexing to read locally
stored email files (.eml, .pst, .mbox formats). Filters can be applied by date range,
sender/receiver, keyword (subject or body), folder, or message size. After import, duplicate
email addresses for the same person are merged using Find & Replace or a Lookup Table.
Types of Email Networks:
• Personal Email Network: Ego-centered — analyzes one person's correspondence. Reveals
social circles, key contacts, forgotten connections.
• Organizational Email Network: Maps communication flow within a company — identifies
boundary spanners, silos, and informal leaders.
• Community Email/Mailing List Network: Analyzes list discussions — identifies experts,
discussion starters, and community health.
Insights from Organizational Email Analysis:
• Identify boundary spanners bridging organizational silos.
• Reveal informal communication structures vs. formal org-charts.
• Detect expertise clusters and isolated employees needing more connections.
• Inform personnel decisions, merger integration, and knowledge management.
Q2. Discuss the structure and analysis of Thread Networks on message boards and
email lists. How do threads create network connections between authors? (Draw and
explain diagram)
Answer:

SNA Answer Key | Page 13

 Thread Networks:
Thread networks emerge from the reply structure of online message boards (forums, Usenet,
Reddit), email lists, and comment sections. Each message and reply creates a directed edge
between authors — when Author A replies to Author B, an edge A→B is added.
Network Construction:
• Vertices: Authors/participants in the discussion.
• Edges: Reply-to relationships — directed from replier to original poster.
• Edge Weight: Number of reply interactions between author pairs.
What Thread Analysis Reveals:
• Discussion starters: Vertices with many incoming edges (many people reply to them).
• Discussion people: Vertices with high in-degree and out-degree — active participants.
• Answer persons: Those who disproportionately provide answers in Q&A forums.
• Lurkers: Participants who read but rarely post — low degree vertices.
• Thread hijackers: Users who derail threads, visible as unexpected reply chains.
(Draw and explain diagram) — Draw a thread tree: original post by A at center, B and C reply
to A (B→A, C→A), D replies to B (D→B), E replies to C (E→C). Show as both a tree and a
network graph.
Q3. Analyze the Twitter network as a social media platform. Describe the types of
connections users create through tweets, replies, mentions, and retweets. How can
NodeXL be used to import and visualize a Twitter conversation network?
Answer:
Twitter Network Connections:
• Reply-to: User responds directly to another's tweet → directed edge from responder to
original poster.
• Mention: User includes @handle in a tweet → directed edge from mentioner to mentioned
user.
• Retweet: User shares another's tweet → directed edge from retweeter to original author.
• Follow: User subscribes to another's tweets → directed edge (separate follower network).
NodeXL Twitter Import Process:
• Open NodeXL > Data > Import > From Twitter Search Network (NodeXL Pro).
• Enter search term (hashtag, keyword, or username).
• Specify number of tweets, time period, and edge types (mentions, replies, retweets).
• NodeXL fetches data via Twitter API and populates the Edges and Vertices worksheets.
• Click Show Graph and apply Fruchterman-Reingold layout.
• Run community detection to identify clusters.
Example — US Senate Twitter Network (Figure 3.1):
The Senate Twitter network showed 100 vertices (senators) connected by mention/reply
edges. Two large clusters (G1, G2) emerged corresponding to the two political parties, with
minimal cross-cluster connections — evidence of political polarization.
Q4. Explain the concept of 'conversation networks' on Twitter. What patterns of
interaction can be identified?
Answer:
Conversation Networks:
A Twitter conversation network captures the reply and mention relationships generated
around a specific topic, hashtag, or event. NodeXL's Twitter importer can collect these
networks, revealing how information flows and who drives the conversation.
Network Patterns (Smith et al. research):
• Polarized Crowd: Two large, separate clusters with minimal inter-cluster connections —
typical of politically divisive topics (e.g., #GunControl). Groups discuss the same topic but
rarely interact with each other.

SNA Answer Key | Page 14

 • Tight Crowd: One dense, highly interconnected cluster — typical of professional
conferences (#AIConf) where attendees know each other.
• Brand Cluster: One hub (brand account) with many spokes (customers mentioning the
brand) but minimal customer-to-customer connections.
• Broadcast Network: One or few major hub accounts sending out information; followers
retweet but rarely reply to each other — typical of news and celebrity accounts.
• Support Network: A hub (company support account) connected to many individual
customers with bidirectional conversations.
• Community Cluster: Multiple dense sub-communities connected by a few bridge accounts.
Q5. What is the role of 'information diffusion' in Twitter networks? How do retweets
and mentions spread information and how are influencers identified?
Answer:
Information Diffusion:
Information diffusion is the process by which content (tweets, links, ideas) spreads through a
network. Twitter's architecture — particularly retweets — makes it an ideal platform for rapid,
viral information spread. A single tweet from a high-follower account can reach millions within
minutes.
Retweets as Diffusion Mechanism:
Each retweet extends the reach of a message to the retweeter's followers. A cascade begins
when many users retweet the same original tweet, creating a tree-like diffusion structure.
Retweet chains can be traced to identify the original source and the path of diffusion.
Mentions as Diffusion:
Mentions draw attention and can introduce content to new audiences. When influential users
mention a topic, it exposes that topic to their follower networks.
Identifying Influencers via SNA Metrics:
• In-degree: High in-degree = many people mention/reply to this user = high audience
engagement.
• PageRank/Eigenvector Centrality: Connected to other important, high-follower accounts.
• Betweenness Centrality: Bridges between different communities — key information brokers.
• Follower Count: Broad potential reach (though reach ≠ engagement).
Q6. Compare and contrast network structures formed by: (i) corporate email, (ii)
academic discussion forum, and (iii) Twitter hashtag conversation.
Answer:
• Corporate Email Network: Directed, weighted. Vertices = employees. Edges weighted by
message volume. Reveals formal/informal hierarchy. Contains organizational silos
(clusters) and boundary spanners. Typically low density — most people communicate
within departments. Privacy-sensitive data requiring ethical handling.
• Academic Discussion Forum: Directed thread network. Vertices = forum members. Edges
from reply chains. Reveals expert/novice distinctions, discussion starters, and answer
persons. Community is defined by topic (e.g., StackOverflow for programming). Public
data accessible for research.
• Twitter Hashtag Conversation: Directed multiplex network (retweets, mentions, replies).
Open, public data. Can scale to millions of users. Often shows polarized crowd or
broadcast patterns. Rapid formation and dissolution around events. Key metrics: retweet
cascade depth, mention in-degree, cluster modularity.
Q7. Describe a case study of analyzing an email archive using SNA. What ethical
considerations arise, and what insights can be discovered?
Answer:
Case Study: Enron Email Archive:
The Enron email dataset (~500,000 emails from ~150 employees released during legal
proceedings) is a landmark SNA case study. Analyzing this archive revealed: key executives

SNA Answer Key | Page 15

 who acted as central communicators; isolated employees unaware of fraud; clusters of
communication around illicit activities; timeline analysis showing abnormal communication
spikes before the bankruptcy.
Insights from Email SNA:
• Identify key influencers and boundary spanners within an organization.
• Detect siloed departments that fail to communicate effectively.
• Track temporal changes (e.g., declining communication in a failing project).
• Identify expertise clusters and up-and-coming leaders.
Ethical Considerations:
• Privacy: Corporate email is legally the employer's property, but employees often assume
some privacy. Transparency about monitoring is ethically required.
• Informed Consent: Employees should be informed about network analysis use of their
communications.
• Anonymization: Pseudonyms protect individuals' identities in research. However, unique
network patterns can re-identify individuals.
• Data Security: Sensitive business information must be protected during analysis.
• Legal Compliance: Subpoenaed email data must be handled within legal frameworks.

UNIT IV – Case Studies: Facebook Networks and WWW

Hyperlink Networks

Part A: 1-Mark Answers

Q1. What type of network is a Facebook friendship network — directed or undirected?

Answer:
A Facebook friendship network is undirected (symmetric/mutual). Both users must agree to
the friendship — if A is friends with B, then B is necessarily friends with A. This is in contrast
to Twitter's follow network which is directed (A can follow B without B following A).
Q2. Define 'hyperlink network' in the context of the World Wide Web.
Answer:
A hyperlink network is a directed network where web pages are vertices and hyperlinks
(clickable URLs from one page to another) are directed edges. The WWW is modeled as a
massive directed graph where in-degree (links pointing to a page) indicates the page's
importance or authority, and out-degree indicates how many pages it references.
Q3. What is the 'six degrees of separation' concept and how does it relate to
Facebook?
Answer:
The 'six degrees of separation' concept (from Stanley Milgram's 1967 experiment) suggests
any two people in the world can be connected through a chain of at most six acquaintances.
Facebook research in 2016 found this had shrunk to an average of only 3.57 intermediaries
among 1.59 billion Facebook users — demonstrating the 'small world' property of large social
networks.
Q4. What is an 'in-link' and an 'out-link' in a WWW hyperlink network?
Answer:

SNA Answer Key | Page 16

 An in-link (backlink) is a hyperlink from another web page pointing TO a given page — it is the
source of the page's authority/importance. An out-link is a hyperlink FROM a given page
pointing to other pages — representing the page's references. In-degree (number of in-links)
is the basis for PageRank and HITS authority scores.
Q5. Define 'echo chamber' in the context of Facebook network analysis.
Answer:
An echo chamber (or filter bubble) is a network phenomenon where users are primarily
exposed to content and opinions that reinforce their existing beliefs, because they are only
connected to like-minded individuals. Facebook's algorithm amplifies this by preferentially
showing content from similar friends. SNA identifies echo chambers as highly clustered
groups with minimal cross-cluster connections.
Q6. What is the PageRank algorithm designed to measure in WWW networks?
Answer:
PageRank measures the relative importance/authority of web pages within the WWW
hyperlink network. A page's PageRank is determined by: (1) number of pages linking to it, (2)
the PageRank of those linking pages, and (3) the number of outgoing links from those pages.
Pages linked by high-PageRank sites receive higher scores — capturing the idea that
important pages are cited by other important pages.
Q7. What is meant by 'triadic closure' in a social network?
Answer:
Triadic closure is the tendency in social networks for two people who share a common friend
to eventually become connected to each other — 'a friend of a friend becomes a friend.' If A-B
and B-C are connected, triadic closure predicts an A-C edge will form. This increases
clustering coefficient and leads to tightly knit community clusters.
Q8. Define 'homophily' in social network analysis.
Answer:
Homophily is the tendency of individuals to form connections with others who are similar to
themselves on key attributes such as age, education, income, ethnicity, political beliefs, and
religion. The phrase 'birds of a feather flock together' captures this principle. Homophily
explains why SNA reveals demographically homogeneous clusters.
Q9. What is a 'fan page' network on Facebook?
Answer:
A fan page network on Facebook consists of public pages created by brands, public figures,
or organizations, with users who 'like' the page. It creates a bimodal affiliation network —
users on one side, pages on the other, with 'like' edges connecting them. This can be
transformed into a user-to-user network (users connected by shared page likes) or a page-to-
page network (pages connected by shared fans).
Q10. What does the term 'web crawl' mean in the context of hyperlink network
analysis?
Answer:
A web crawl is an automated process where a program (web crawler/spider) starts at seed
URLs and systematically follows hyperlinks to discover and collect data from web pages. In
hyperlink network analysis, crawlers map the structure of the WWW by recording which pages
link to which other pages, building the edge list for hyperlink network analysis.

Part B: 10-Mark Answers

SNA Answer Key | Page 17

Q1. Describe the structure of Facebook friendship networks. What types of network
connections are created on Facebook? How can NodeXL be used to visualize these
networks?
Answer:
Facebook Network Structure:
Facebook generates multiple overlapping network types. The primary network is the
friendship network — an undirected, unweighted network where vertices are users and edges
represent mutual friendship agreements.
Types of Facebook Network Connections:
• Friend Network: Undirected, mutual. User A and User B both agree to connect.
• Page/Group Affiliation Network: Bimodal — users connected to pages they 'like' or groups
they join.
• Co-tag Network: Users tagged together in the same photo are connected.
• Interaction Network: Directed, weighted — users who comment on, like, or share each
other's posts (captures actual engagement, not just declared friendship).
• Event Attendance Network: Users who attend the same Facebook events are connected.
NodeXL Facebook Import (Legacy):
Earlier versions of NodeXL Pro included Facebook importers. The data was imported as an
edge list, then visualized using force-directed layouts. Vertex size could map to friend count;
color to group membership; community detection revealed social circles.
Key Findings from Facebook SNA:
Research consistently reveals tight clusters (friend groups, family, coworkers), low inter-
cluster density (weak ties between groups), and key bridge users who connect otherwise
separate social circles. The average Facebook user has 3.57 degrees of separation from any
other user.
Q2. Explain the concept of WWW Hyperlink Networks. How are web pages and
hyperlinks modeled as vertices and edges? (Draw and explain diagram)
Answer:
WWW as a Network:
The World Wide Web is modeled as a massive directed graph. Every web page is a vertex.
Every hyperlink from Page A to Page B is a directed edge: A → B. The edge means A 'points
to' or 'endorses' B.
Structural Properties:
• In-degree: Number of pages linking TO a given page. High in-degree = authoritative/popular
page (basis of PageRank).
• Out-degree: Number of links FROM a page to others. High out-degree = hub/directory-type
page.
• Scale-free Property: WWW follows a power-law degree distribution — a few massive hubs
(Google, Wikipedia, Amazon) have enormous in-degree while most pages have few links.
• Small-world Property: Despite billions of pages, any two pages can typically be reached in a
small number of clicks.
HITS Algorithm:
Kleinberg's HITS (Hyperlink-Induced Topic Search) algorithm identified two roles in web
hyperlink networks: Hubs (pages with many outgoing links to quality pages) and Authorities
(pages with many incoming links from quality hubs). This later inspired Google's PageRank.
(Draw and explain diagram) — Draw 5 web page nodes. Show directed arrows as hyperlinks.
Label one page as 'High Authority' (many in-links) and one as 'Hub' (many out-links).
Q3. Discuss the phenomenon of 'echo chambers' and 'filter bubbles' in Facebook
networks. How do SNA techniques help identify these? What are the societal
implications?
Answer:

SNA Answer Key | Page 18

 Echo Chambers:
An echo chamber is a clustered subnetwork where users primarily interact with others who
share their beliefs, reinforcing existing viewpoints while excluding contradictory information.
Facebook's friendship selection (homophily) combined with its algorithmic content ranking
(showing you more content your friends engage with) creates these chambers naturally.
Filter Bubbles:
Coined by Eli Pariser, a filter bubble is the state of intellectual isolation created by algorithmic
personalization. Facebook shows you posts that match your prior engagement patterns,
effectively filtering out diverse perspectives without your conscious awareness.
SNA Detection Methods:
• Modularity Score: High modularity in community detection indicates strong clustering — a
sign of potential echo chambers.
• Cross-cluster Edge Ratio: Low ratio of inter-cluster to intra-cluster edges indicates isolated
communities.
• Content Diversity Analysis: Combined with text analysis, SNA can show whether clusters
discuss the same topics differently or avoid certain topics entirely.
• Bridge Identification: Betweenness centrality identifies rare bridge users connecting clusters
— potential diversity introducers.
Societal Implications:
• Political polarization: Each side cannot understand the other's perspective.
• Misinformation spread: False information circulates rapidly within clusters unchallenged.
• Radicalization: Extreme views are reinforced without counterarguments.
• Democratic discourse erosion: Shared reality breaks down across communities.
Q4. Explain how the PageRank algorithm uses hyperlink structure to rank web pages.
Compare PageRank with degree and betweenness centrality.
Answer:
PageRank Algorithm:
PageRank (developed by Brin & Page at Google, 1998) assigns each web page a score
based on the principle that a page is important if important pages link to it. Formula considers:
(1) Number of in-links to the page; (2) PageRank scores of the linking pages; (3) Number of
outgoing links from the linking pages (link propensity).
Intuitively: A single link from the New York Times website is worth far more than 100 links
from obscure personal blogs — because the NYT has high PageRank.
Comparison:
• Degree Centrality: Simple count of connections. Does not distinguish between linking to a
hub vs. a peripheral page. Doesn't consider the quality of connections.
• Betweenness Centrality: Measures bridge/broker role — how often the page lies on the
shortest path between other pages. Computationally expensive for the full WWW scale.
• PageRank/Eigenvector Centrality: Iterative, quality-weighted measure. A page inherits
importance from the importance of its linkers. Scales to billions of pages. Most appropriate
for directed hyperlink networks where the quality and authority of connections matters.
Q5. Analyze the differences between a unimodal Facebook friend network and a
multimodal Facebook network. How is a bimodal affiliation network transformed for
analysis?
Answer:
Unimodal Facebook Friend Network:
Contains one vertex type (users) and one edge type (friendship). All connections are
symmetric. Analysis reveals friendship clusters, bridge users, and social capital structures.
Standard network metrics (degree, betweenness, clustering coefficient) apply directly.
Multimodal Facebook Network:

SNA Answer Key | Page 19

 Contains multiple vertex types: Users, Pages, Groups, Events, Posts. Edges cross between
types: User 'likes' Page, User 'joins' Group, User 'attends' Event, User 'comments on' Post.
This creates a rich, complex network but requires special handling since standard metrics
assume unimodal networks.
Bimodal Affiliation Network Transformation:
A bimodal network (e.g., Users ↔ Pages) has two vertex types and only cross-type edges. To
apply standard SNA metrics, it must be transformed into two unimodal networks:
• User-to-User Network: Two users are connected if they both 'like' the same page —
connected through their shared page affiliation. Edge weight = number of shared pages.
• Page-to-Page Network: Two pages are connected if the same user 'likes' both. Edge weight
= number of shared fans.
This transformation is the basis of recommender systems — 'Pages people who liked X also
liked.'
Q6. Describe a case study of visualizing and interpreting a Facebook political network
during an election. What clusters, key influencers, and information pathways would
SNA reveal? (Draw and explain diagram)
Answer:
Case Study: Facebook Political Network Analysis:
Researchers studying political discussions on Facebook (e.g., during elections) collect
interaction networks — users connected by comments, shares, and likes on political posts.
SNA reveals the following:
Clusters:
Community detection typically reveals strong partisan clusters — liberal-leaning users interact
primarily with each other; conservative-leaning users form a separate cluster. Cross-partisan
connections (bridge users) are rare. This maps directly to real-world political division.
Key Influencers:
• High in-degree: Political celebrities, news pages that receive many interactions.
• High betweenness centrality: Bridge users who cross partisan lines — often journalists,
centrists, or bipartisan organizations.
• High PageRank: Pages or users whose content is shared by other high-engagement
accounts.
Information Pathways:
Within clusters, misinformation spreads rapidly (viral sharing without fact-checking). Between
clusters, information flows only through rare bridge users. SNA can identify which bridge
users are most effective at cross-cluster communication.
(Draw and explain diagram) — Draw two dense clusters (Blue = liberal, Red = conservative)
with minimal cross-edges. Highlight bridge users in green. Show information flow arrows
within and between clusters.
Q7. Compare the network structures of Facebook (social) and WWW (hyperlink).
Discuss metrics such as density, diameter, clustering coefficient, and scale-free
properties.
Answer:
• Network Type: Facebook friendship = undirected, symmetric. WWW hyperlinks = directed,
asymmetric.
• Density: Both are extremely sparse relative to possible connections. Facebook: Most users
have a few hundred friends out of billions of possible friends. WWW: Most pages link to
dozens out of billions of possible pages.
• Diameter: Both exhibit small-world properties. Facebook: avg. 3.57 degrees of separation.
WWW: researchers estimate ~19 clicks connect any two pages on the indexed web.

SNA Answer Key | Page 20

 • Clustering Coefficient: Facebook has a high clustering coefficient — friends of friends tend
to become friends (triadic closure). WWW has variable clustering — topic-specific web
communities (e.g., academic sites, news sites) have high local clustering.
• Scale-Free Properties: Both follow power-law degree distributions. Facebook: A few
celebrity accounts have millions of friends; most users have hundreds. WWW: A few
pages (Google, Wikipedia) have billions of in-links; most pages have very few. This scale-
free property means the network is robust to random node removal but vulnerable to
targeted attacks on hubs.

UNIT V – Case Studies: YouTube Networks and Wiki

Networks

Part A: 1-Mark Answers

Q1. What is a 'content interaction network' on YouTube?

Answer:
A content interaction network on YouTube is a network formed by the interactions users have
with video content — comments, likes, replies to comments, and video responses. Vertices
represent users or videos; edges represent comment, reply, or shared viewership
connections. These networks reveal patterns of engagement around specific content.
Q2. Define 'prominence' in the context of YouTube network analysis.
Answer:
Prominence in YouTube network analysis refers to the degree to which a user or channel
stands out relative to others in the network — measured by metrics such as high view counts,
subscriber count, in-degree (many users commenting on or referencing them), or high
PageRank/eigenvector centrality (connected to other prominent channels or users).
Q3. What is a 'wiki' and what collaborative principle underlies its design?
Answer:
A wiki is a web-based collaborative platform that allows multiple users to create, edit, and link
pages using a simplified markup language. The underlying principle is open, collaborative
knowledge creation — anyone (registered or anonymous) can contribute and edit content.
Wikipedia is the world's largest wiki, with millions of articles co-authored by volunteers. The
MediaWiki platform supports structured text, auto-generated tables of contents, and version
history.
Q4. What is a 'co-edit network' in wiki analysis?
Answer:
A co-edit network is a network constructed from Wikipedia editing behavior — two editors are
connected if they have both edited the same article. Vertices = editors; edges = co-authorship
of articles. Edge weight = number of articles co-edited. This unimodal editor-to-editor network
is derived by transforming the bimodal editor-to-article affiliation network.
Q5. Define 'talk page network' in Wikipedia.
Answer:
A talk page is a discussion page associated with each Wikipedia article where editors discuss
proposed changes, dispute resolution, and editorial decisions. A talk page network connects

SNA Answer Key | Page 21

 editors who participate in the same article's talk page discussion. It represents editorial
coordination and conflict resolution — edges form when editors post on the same talk page.
Q6. What is meant by 'contrasting patterns of content interaction' on YouTube?
Answer:
Contrasting patterns of content interaction on YouTube refers to the observation that different
types of content (news videos, entertainment, tutorials, political content) generate
fundamentally different network structures. News generates broad, shallow comment
networks; entertainment generates high-engagement, dense reply trees; political content
generates polarized clusters with hostile cross-group interactions — similar to the polarized
crowd pattern on Twitter.
Q7. What type of network is formed when YouTube users comment on each other's
videos?
Answer:
When users comment on each other's videos, a directed bimodal network is formed — users
(vertex type 1) connected to videos (vertex type 2). This can be further transformed: users
who comment on the same video are indirectly connected (co-commenter network), or videos
that share the same commenters are connected (co-viewed/co-commented video network).
Q8. Define 'Connections of Creativity and Collaboration' in the context of wiki
networks.
Answer:
'Connections of Creativity and Collaboration' refers to the networks formed when contributors
collaboratively create and curate knowledge in wiki environments. These connections reveal
the social structure of knowledge production — who collaborates with whom, which editors
are central to knowledge domains, and how creative contributions flow through the
community.
Q9. What is an 'affiliation network' in the context of Wikipedia editors and articles?
Answer:
In Wikipedia, an affiliation network is a bimodal network where one vertex type = editors and
the other vertex type = articles. An edge (affiliation) exists when an editor has made at least
one edit to an article. No direct editor-to-editor or article-to-article edges exist in the bimodal
form. This affiliation network is the raw material for recommender systems and collaborative
filtering.
Q10. Name one key metric used to identify prominent contributors in a YouTube or
Wiki network.
Answer:
In-degree centrality is a key metric — editors or users with many incoming connections (many
replies to their comments, or many co-editors who also edited their articles) are identified as
prominent contributors. PageRank is also used — contributors who are connected to other
high-activity contributors receive higher PageRank scores, identifying them as core
community members.

Part B: 10-Mark Answers

Q1. Describe the types of networks that emerge from YouTube interactions. How can
these networks be visualized using NodeXL?
Answer:
Types of YouTube Networks:

SNA Answer Key | Page 22

 • User Reply Network: Directed network of replies in YouTube comment sections. User A
replies to User B's comment → directed edge A→B. Reveals conversation hierarchies,
dominant voices, and trolls.
• Video Co-view Network (Bimodal): Users connected to videos they have
watched/commented on. Transformed into video-to-video network (videos with shared
audiences) or user-to-user network (users with shared video interests).
• Channel Subscription Network: Users connected to channels they subscribe to — a bimodal
network of users and channels.
• Video Response Network: Directed edges from response videos to original videos — less
common after YouTube removed the feature, but historically formed citation-like networks.
• Related Video Network: YouTube's recommendation system connects videos — creating a
directed network showing which videos YouTube's algorithm recommends after viewing
another.
NodeXL Visualization Process:
• Import YouTube data via NodeXL Pro's YouTube Network importer.
• Specify search terms or video IDs to collect comment/reply networks.
• The Edges worksheet populates with commenter-to-commenter reply pairs.
• Apply Fruchterman-Reingold layout and community detection.
• Map vertex size to in-degree (comment engagement) and color to community group.
Q2. Explain the concept of 'Contrasting Patterns of Content Interaction' on YouTube.
What different roles do users play? How can SNA metrics identify prominent users?
Answer:
Contrasting Patterns:
Different content categories on YouTube generate structurally distinct interaction networks,
reflecting the different social dynamics of their viewer communities.
• Entertainment/Music Content: High view counts, shallow comment trees (many independent
top-level comments, few replies). Star-shaped networks centered on popular videos.
• News/Political Content: Polarized networks — viewers with opposing viewpoints form
separate clusters with hostile bridge interactions. High controversy generates more
replies.
• Tutorial/Educational Content: Q&A-style networks — students ask questions (replies
directed to creator), creator responds. Similar to 'answer person' pattern.
• Community/Niche Content: Dense, reciprocal networks — community members know each
other, reply frequently, form tight clusters.
User Roles on YouTube:
• Broadcasters/Creators: Few outgoing interactions, very many incoming (replies, likes). High
in-degree.
• Active Commenters/Discussion People: High in-degree and out-degree — participate
heavily.
• Trolls/Provocateurs: High betweenness — bridge between hostile clusters.
• Lurkers/Viewers: Rarely comment — not visible in comment networks.
SNA Metrics for Prominence:
• In-degree: Number of replies received.
• PageRank: Importance weighted by the importance of who replies to you.
• Betweenness: Bridge between discussion subgroups.
Q3. Discuss how Wiki networks represent 'Connections of Creativity and
Collaboration.' Explain how editor-to-article affiliation networks can be transformed
into editor-to-editor collaboration networks. (Draw and explain diagram)
Answer:
Wiki Networks as Creative Collaboration:

SNA Answer Key | Page 23

 Wikipedia represents one of history's largest collaborative knowledge creation projects —
millions of volunteer editors collaboratively author, edit, and curate articles. The network
structure of this collaboration reveals the social organization of knowledge production: who
collaborates with whom, which editors are domain experts, and how knowledge communities
are structured.
Bimodal Affiliation Network:
The raw data forms a bimodal network: Vertices = {Editors} ∪ {Articles}. Edges = 'Editor E has
edited Article A.' No direct editor-to-editor edges exist in this bimodal form.
Transformation to Unimodal Networks:
• Editor-to-Editor (Co-edit) Network: Two editors are connected if they both edited the same
article. Edge weight = number of shared article edits. Derek and Marc, who co-edit many
wiki pages, would have a strong connection.
• Article-to-Article Network: Two articles are connected if many shared editors contributed to
both. Articles in the same subject domain (e.g., Machine Learning and Neural Networks)
will be strongly connected through shared expert editors.
(Draw and explain diagram) — Draw bimodal graph: Editor nodes E1,E2,E3 on left; Article
nodes A1,A2,A3 on right. Show edit affiliations as edges. Then draw resulting E1-E2 and E2-
E3 unimodal edges when transformed.
Q4. Describe a case study of analyzing a Wikipedia co-authorship network. What does
the structure reveal about knowledge production, community roles, and editorial
patterns?
Answer:
Case Study: Wikipedia Co-authorship Network
Researchers construct Wikipedia co-authorship networks by analyzing edit histories of article
collections (e.g., all articles tagged under 'Artificial Intelligence'). The resulting editor-to-editor
network reveals the social structure of knowledge production.
Findings from Network Analysis:
• Core-Periphery Structure: A small core of highly active editors contribute the majority of
edits (power-law distribution). These core editors are strongly interconnected. A large
periphery of occasional contributors make fewer edits with weaker connections.
• Domain Clustering: Editors tend to cluster around subject areas — biologists cluster around
biology articles, technologists around tech articles. These clusters are connected by a few
generalist bridge editors.
• Social Roles: 'Gnomes' (minor correction editors), 'Vandal fighters' (reverting bad edits),
'Substantive contributors' (adding new content), 'Talk page negotiators' (resolving
disputes). Each role has a distinctive network signature (degree, betweenness, clustering
coefficient).
• Edit War Detection: Articles with frequent sequential reverts between two editors show in the
network as high-weight, bidirectional 'conflict edges' — SNA can flag articles prone to edit
wars.
• Quality Correlation: Articles with more diverse, large contributor communities tend to have
higher quality ratings — network diversity predicts article quality.
Q5. Compare and contrast YouTube networks and Wikipedia networks in terms of
network type, user roles, edge types, community structure, and what SNA can reveal.
Answer:
• Network Type: YouTube generates directed bimodal (user-video) or directed unimodal
(user-user reply) networks. Wikipedia generates bimodal (editor-article) affiliation
networks, transformed to unimodal co-edit networks.
• User Roles: YouTube: broadcasters, commenters, trolls, lurkers. Wikipedia: gnomes, vandal
fighters, substantive contributors, talk page negotiators, administrators.

SNA Answer Key | Page 24

 • Edge Types: YouTube: reply-to (directed), subscription (directed), co-view (undirected
derived). Wikipedia: co-edit (undirected derived from affiliation), talk page co-participation
(undirected).
• Community Structure: YouTube shows polarized crowds (political/news content) or tight
niche communities (hobbyist channels). Wikipedia shows domain-clustered communities
(science editors, history editors) with core-periphery structure.
• SNA Insights: YouTube: identifies influencers, viral content, toxic communities. Wikipedia:
identifies knowledge gaps, article quality predictors, community health, expert editors, and
edit conflict zones.
• Scale: YouTube networks scale to millions of users per video. Wikipedia co-authorship
networks are typically analyzed at article-collection or domain level (thousands of editors).
Q6. Discuss social roles in online content communities using YouTube and Wiki as
examples. How do SNA metrics identify 'answer persons,' 'discussion people,' and
'peripheral lurkers'?
Answer:
Social Roles in Online Communities:
Studies by Welser, Gleave, and Smith demonstrated that contributors in online communities
create distinctive network patterns reflecting their roles — evidence of behavioral
specialization. Different roles leave different 'network signatures.'
Answer Person:
Disproportionately provides answers to questions (especially in Q&A forums and YouTube
tutorials). Network signature: High out-degree (many replies sent), low in-degree (fewer
people reply back), connections spread across many different conversation threads rather
than concentrated. In Wikipedia: the 'gnome' who fixes minor errors across many articles.
Discussion Person:
Engages in extended exchanges — participates in long conversation threads. Network
signature: High in-degree AND high out-degree, high clustering coefficient (densely
connected within a community cluster), long path lengths within threads.
Discussion Starter:
Starts threads that others engage with heavily. High betweenness centrality — acts as the
origin point for many conversations. High in-degree from diverse reply sources.
Influential Person:
Well-connected to others who are more highly connected. High eigenvector centrality /
PageRank. May not have highest raw degree but connects to hubs.
Peripheral Lurker:
Reads content but rarely posts. Very low degree (near isolate). In wiki networks: edits rarely,
perhaps only making minor corrections. In YouTube: watches videos but never comments —
invisible in comment networks.
Boundary Spanner:
Bridges between unconnected subgroups. High betweenness centrality, low clustering
coefficient (connects people who don't know each other). Critical for cross-community
information flow.
Q7. Synthesize your understanding of all five units: Compare the network structures,
analytical tools, and insights from analyzing Email, Twitter, Facebook, WWW,
YouTube, and Wiki networks using SNA.
Answer:
Comparative Overview:
• Email Networks: Directed, weighted. Most privacy-sensitive. Reveals organizational
communication structure, informal hierarchies, boundary spanners, and temporal
relationship patterns. Data from personal/corporate archives.

SNA Answer Key | Page 25

• Twitter Networks: Directed, multiplex (reply, mention, retweet). Public data. Rapidly forming
around events/hashtags. Reveals polarization patterns, information diffusion cascades,
influencers, and conversation structure types (broadcast, tight crowd, polarized crowd).
• Facebook Networks: Undirected friendship, multimodal (users/pages/groups). Semi-public.
Reveals echo chambers, filter bubbles, homophily, and social capital structures. Six
degrees of separation demonstrated.
• WWW Hyperlink Networks: Directed, massive scale. Reveals authority/hub structure, scale-
free topology, small-world properties. Foundation for PageRank and search engine
algorithms.
• YouTube Networks: Directed (bimodal + unimodal). Reveals content interaction patterns,
broadcaster vs. community content, viral diffusion, and social roles in media consumption.
• Wiki Networks: Bimodal affiliation (editor-article), transformed to co-edit networks. Reveals
collaborative knowledge production, community structure, editorial roles, and content
quality predictors.
Common Analytical Tools:
All six network types are analyzed using NodeXL for visualization; Fruchterman-Reingold and
Harel-Koren for layout; Clauset-Newman-Moore for community detection; and metrics: degree
centrality, betweenness, closeness, eigenvector/PageRank, density, clustering coefficient.
Common Themes:
• Scale-free degree distributions (power law) across all platforms.
• Small-world properties — short average path lengths despite low density.
• Community structure — clustering around shared interests, beliefs, or topics.
• Social role specialization — answer persons, broadcasters, lurkers exist across all
platforms.
• Information diffusion follows network structure — hubs accelerate spread; bridges enable
cross-community flow.
Social Network Analysis provides the unified theoretical framework and quantitative tools to
study all these platforms with consistent, comparable methods — enabling cross-platform
insights into human communication, collaboration, and community formation.

SNA Answer Key | Page 26