• Query successful

Here's a breakdown of the answers to Question 1 from an exam perspective, keeping in

mind the marks allocated:

Question 1 (5x5=25 marks)

General Approach for 5-mark questions:

• Definition: Start with a clear and concise definition.

• Explanation/Elaboration: Expand on the definition, providing more details.

• Example (if applicable): Give a relevant example to illustrate the concept.

• Key Characteristics/Components (if applicable): List important features or parts.

• Conciseness: Be to the point; avoid unnecessary jargon or lengthy descriptions.

(a) What is a Sociogram? Explain it with help of example.

• Definition: A sociogram is a graphic representation or map of social links that a

person has. It's a visual tool used to plot and analyze the relationships among
individuals or groups within a social network.

• Explanation: In a sociogram, individuals are typically represented as nodes (or

points/circles), and the relationships between them are represented as lines (or
edges/arrows). The type of relationship (e.g., friendship, communication,
collaboration) can be indicated by different line types or directions. It helps visualize
patterns of interaction, identify central figures, cliques, and isolates within a network.

• Example: Imagine a group of five friends: Alice, Bob, Carol, David, and Emily.

o Alice is friends with Bob and Carol.

o Bob is friends with Alice and David.

o Carol is friends with Alice and Emily.

o David is friends with Bob.

o Emily is friends with Carol.

o A sociogram would show five nodes (one for each person). Lines would
connect Alice to Bob, Alice to Carol, Bob to David, Carol to Emily. This visual
representation immediately shows who is connected to whom and can reveal
patterns like Alice being a central figure.

(b) What are the key components of social network analysis? What is called prominence?
 • Key Components of Social Network Analysis (SNA):

1. Actors/Nodes/Vertices: These are the individual entities within the network

(e.g., people, organizations, websites, molecules).

2. Ties/Edges/Links/Relationships: These are the connections or relationships

between the actors (e.g., friendship, communication, trade, citation). Ties can
be directed (e.g., A follows B) or undirected (e.g., A is friends with B). They
can also have weights (e.g., frequency of interaction).

3. Network Structure: This refers to the overall pattern of connections within

the network, including density, centralization, cliques, and bridges.

4. Attributes: These are characteristics of the actors or ties (e.g., age, gender,
type of relationship, strength of connection).

• Prominence: In social network analysis, "prominence" generally refers to the

importance or influence of a node within the network. It's a broad concept often
quantified by various centrality measures. A prominent node is one that holds a
significant position, often having many connections, being strategically located, or
having control over information flow. It indicates a node's visibility and potential
impact within the network.

(c) Define Transitivity matrix of social network graphs.

• Definition: In the context of social network graphs, a transitivity matrix (or often
referred to as an adjacency matrix when discussing paths and connections) is used to
represent the relationships between nodes and to understand transitive properties.
While a "transitivity matrix" isn't a standard independent matrix name, the concept
is usually embedded within the adjacency matrix and its powers.

• Explanation (using Adjacency Matrix): An adjacency matrix (A) for a graph with 'n'
nodes is an n×n matrix where Aij=1 if there is a direct tie from node i to node j, and
Aij=0 otherwise.

o Transitivity: The concept of transitivity in networks is about whether "friends

of my friends are also my friends." If A is connected to B, and B is connected
to C, is A also connected to C?

o To understand transitivity mathematically using matrices, we often look at

powers of the adjacency matrix. The element (Ak)ij of the k-th power of the
adjacency matrix indicates the number of paths of length k between node i
and node j.

o For transitivity, we are specifically interested in paths of length 2. If (A2)ij>0, it

means there is at least one path of length 2 between i and j. This path goes
from i to some intermediate node k, and then from k to j. The concept of
 transitivity often involves examining if a direct tie exists (or is likely to exist)
when a path of length 2 is present. For instance, in undirected networks, if
A−B and B−C exist, a transitive relationship would imply A−C also exists,
forming a triangle. This is often measured by the clustering coefficient, which
uses properties derived from the adjacency matrix and its squares.

(d) Define clique and cutip with help of example.

• Clique:

o Definition: In a graph, a clique is a subset of vertices such that every pair of

distinct vertices in the subset is connected by an edge. In simpler terms, it's a
maximally connected subgraph where all members are directly connected to
all other members within that subset.

o Explanation: Cliques represent very cohesive and tightly-knit groups within a

larger network. They are often found in friendship groups where everyone is
friends with everyone else.

o Example: Consider a network with nodes A, B, C, D, E. If A is connected to B,

C, and D; B is connected to A, C, and D; C is connected to A, B, and D; and D is
connected to A, B, and C. Then {A, B, C, D} forms a clique because every
person in this group is directly connected to every other person in the group.

• Cut-Point (often referred to as Articulation Point):

o Definition: A cut-point (or articulation point) in a connected graph is a vertex

whose removal (along with all incident edges) would disconnect the graph or
increase the number of connected components.

o Explanation: Cut-points are crucial nodes in a network because they act as

bridges. If a cut-point is removed, communication or flow between different
parts of the network might be severely disrupted. They represent
vulnerabilities or critical junctions.

o Example: Imagine a network representing cities and roads. If City X is the only
road connection between a group of cities (Group A) and another group of
cities (Group B), then City X is a cut-point. If City X is removed (e.g., road
closed), Group A and Group B would no longer be connected by road.

(e) Define network positions.

• Definition: Network positions refer to the structural location or role of an actor

within a social network. These positions are determined by the pattern of
connections an actor has with other actors in the network, rather than their
individual attributes.
 • Explanation: Actors occupying similar network positions tend to have similar
patterns of relationships, even if they are not directly connected. Analyzing network
positions helps understand the roles individuals play (e.g., central connector, bridge,
periphery), how they access information, or how they are influenced by others. Key
concepts related to network positions include:

o Centrality: Measures like degree, betweenness, closeness, and eigenvector

centrality quantify different aspects of a node's importance or influence
based on its connections.

o Equivalence: Concepts like structural equivalence (nodes having identical ties

to identical others) and regular equivalence (nodes having similar patterns of
ties to similar others) categorize nodes into positions.

o Roles: Based on their positions, actors can be identified as leaders,

gatekeepers, liaisons, isolates, etc.

• Importance: Understanding network positions is vital for predicting behavior,

understanding influence, and identifying critical points in a network.

(f) Define the term Triad census.

• Definition: A triad census is a method used in social network analysis to classify and
count all possible configurations of three nodes (a "triad") within a directed network.

• Explanation: For any three nodes, there are 16 possible unique configurations of
directed ties between them. These configurations range from three isolated nodes
(no ties) to a fully connected triangle where all nodes have directed ties to each
other (e.g., A->B, B->C, C->A, A->C, C->B, B->A). The triad census involves
systematically identifying each distinct triad type and counting its occurrences in a
given network.

• Purpose: The distribution of these 16 triad types provides insights into the structural
properties of the network, including transitivity, reciprocity, hierarchy, and the
presence of specific local network patterns. For example, a network with many cyclic
triads (like A->B->C->A) might indicate more balanced power or reciprocal
relationships, while a network with many hierarchical triads (like A->B, B->C, A->C but
no C->A) might suggest power imbalances or strict ordering. It's a fundamental tool
for understanding local network structure and comparing different networks.

(g) What is an example of a small-world network?

• Definition (Implicit): A small-world network is a type of graph that has both a high
clustering coefficient (like regular networks, meaning nodes tend to be
interconnected) and a short average path length (like random networks, meaning any
 two nodes can be reached from each other in a small number of steps). It exhibits a
balance between local connectivity and global reachability.

• Example:

o The World Wide Web: While a massive and complex network, the
connections between web pages (through hyperlinks) exhibit small-world
properties. You can often navigate from one page to almost any other page in
a relatively small number of clicks. You also find that pages linking to each
other (e.g., pages within the same website or related topics) tend to form
tight clusters.

o Social Networks (e.g., Facebook, LinkedIn): This is a classic example. Your

friends on Facebook are likely to be friends with each other (high clustering),
and you can reach almost anyone on the platform through a "friend of a
friend" chain in very few steps (low average path length, famously known as
"six degrees of separation"). You are closely connected to your immediate
social circle, but through a few long-range connections, you can access distant
parts of the network.

o Scientific Collaboration Networks: Researchers who co-author papers form a

network. Scientists tend to collaborate with others in their immediate field
(clustering), but through a few collaborations with researchers in other fields,
the entire scientific community becomes surprisingly interconnected.

question 2

I can provide an answer to Question 2. What is the question?

ques 2 a and b part

What is question 2, parts a and b?

checkin the image

Certainly! Let's break down Question 2, parts (a) and (b), from the provided image, keeping
the exam point of view and marks in mind (5 marks for (a), 7.5 for (b)).

Question 2

(a) What is Sociometry? List any three goals of it. (5 marks)

• What is Sociometry?
 o Definition: Sociometry is a quantitative method for measuring social
relationships. It's a research technique used to chart the relationships
between people within a group, often aiming to identify patterns of
attraction, repulsion, or indifference among group members. It specifically
analyzes group dynamics, leadership structures, and communication flows.

o Key Concept: It often involves asking group members to specify whom they
would like to work with, play with, sit near, or who they dislike, etc., from the
other members of the group. The responses are then compiled, typically into
a sociomatrix or visualized as a sociogram.

• Three Goals of Sociometry:

1. To understand Group Structure and Dynamics: A primary goal is to map out the
informal social organization of a group, revealing cliques, isolates, stars (popular members),
and bridges (members connecting different subgroups). This helps in understanding how a
group functions beyond its formal structure.

2. To identify Leaders and Influencers: By analyzing who receives the most choices
(e.g., for collaboration or friendship), sociometry can identify individuals who are central,
popular, or influential within the group, even if they don't hold formal leadership positions.

3. To improve Group Cohesion and Communication: By diagnosing issues like social

isolation, unreciprocated choices, or communication bottlenecks, sociometry can provide
insights for interventions aimed at improving group morale, fostering better communication
channels, and enhancing overall group effectiveness.

4. (Bonus/Alternative) To facilitate Group Formation or Intervention: In practical

settings (e.g., classrooms, workplaces), sociometry can be used to form more effective
teams, place individuals in roles where they are more likely to thrive, or design interventions
to help integrate isolated individuals.

(b) What is network data analysis? How does network analytics work? (7.5 marks)

• What is Network Data Analysis (or Network Analytics)?

o Definition: Network data analysis (or network analytics) is a field that uses
methods from graph theory, statistics, and computer science to study,
visualize, and understand relationships and interactions between entities
(nodes/actors) in a network. It focuses on the structure of relationships rather
than just the attributes of individual entities.

o Scope: It involves collecting, processing, and analyzing data where the

fundamental unit of observation is a relationship or connection between two
 or more entities. This can include social connections, communication
patterns, biological interactions, transportation routes, information flows, etc.

o Goal: The primary goal is to uncover hidden patterns, identify key players,
predict behaviors, and understand the dynamics of complex systems by
examining the connections within them.

• How does Network Analytics Work? Network analytics typically follows a systematic
process involving several key steps:

1. Data Collection:

▪ Identifying the entities (nodes) and the relationships (edges) between

them.

▪ Sources can be varied: surveys (e.g., sociometric questionnaires),

digital records (e.g., email logs, social media interactions, transaction
data), scientific publications (co-authorship), biological data (protein
interactions), etc.

▪ The data needs to be structured, often as an adjacency list or an

adjacency matrix, where rows and columns represent nodes and
entries indicate the presence or absence (or weight) of a connection.

2. Network Modeling/Representation:

▪ Representing the collected data as a graph where nodes are entities

and edges are relationships.

▪ Deciding on the type of graph: directed vs. undirected, weighted vs.

unweighted, bipartite, etc., based on the nature of the relationships.

3. Network Measurement and Metrics:

▪ Calculating various quantitative metrics to describe the network's

properties and the position of individual nodes.

▪ Node-level metrics:

▪ Centrality Measures: (e.g., Degree Centrality: number of direct

connections; Betweenness Centrality: how often a node lies on
the shortest path between others; Closeness Centrality: how
quickly a node can reach other nodes; Eigenvector Centrality:
influence based on connections to other influential nodes).

▪ Clustering Coefficient: Measures how interconnected a node's

neighbors are.

▪ Network-level metrics:
 ▪ Density: Proportion of actual connections to possible
connections.

▪ Average Path Length: Average number of steps between any

two nodes.

▪ Components: Identification of connected subgraphs.

▪ Community Detection: Algorithms to identify tightly knit

groups or clusters (communities) within the network.

4. Network Visualization:

▪ Creating visual representations of the network (sociograms, force-

directed layouts, etc.) to help understand complex structures, identify
patterns, and communicate findings effectively. Tools like Gephi,
Cytoscape, or networkx in Python are commonly used.

5. Interpretation and Inference:

▪ Drawing meaningful conclusions from the metrics and visualizations.

▪ Answering research questions, identifying key individuals or groups,

predicting behaviors, uncovering hidden structures, and making data-
driven decisions based on the network insights. For example,
identifying influential nodes for targeted interventions, finding
bottlenecks in communication flows, or understanding the spread of
information/disease.

This systematic approach allows network analytics to move beyond simple descriptive
statistics to understand the underlying relational structure that drives phenomena in various
domains.

same for 3

Let's address Question 3 from the image, keeping the exam context and marks in mind.

Question 3

(a) Explain the Structural Balance Theory with help of example. (5 marks)

• Structural Balance Theory (SBT):

o Origin & Core Idea: Proposed by Fritz Heider in psychology and later
formalized in graph theory by Cartwright and Harary, Structural Balance
Theory (SBT) deals with the dynamics of signed social networks. It posits that
 people strive for psychological consistency or "balance" in their relationships,
particularly within a triad (a group of three individuals or entities).

o Signed Relationships: The theory applies to networks where ties between

actors can be either positive (+) (e.g., friendship, liking, trust, alliance) or
negative (-) (e.g., enmity, disliking, distrust, rivalry).

o Principle of Balance: SBT suggests that certain configurations of relationships

within a triad are "balanced" (stable and comfortable), while others are
"unbalanced" (unstable and prone to change). The network will tend to
evolve towards a balanced state.

o Rules for Balanced Triads: A triad is considered balanced if the product of the
signs of its three relationships is positive. This leads to two fundamental types
of balanced triads:

1. Three positive ties (+++): A likes B, B likes C, and A likes C. (Friend of a

friend is a friend). This is stable and harmonious.

2. One positive tie and two negative ties (+−−): A likes B, B dislikes C,
and A dislikes C. (Friend of an enemy is an enemy, or enemy of an
enemy is a friend). This is also stable; for example, if A and B are
friends, and B and C are enemies, it's consistent for A to also be an
enemy of C. Or, if A and B are enemies of C, it's consistent for A and B
to be friends.

o Unbalanced Triads: If the product of the signs is negative, the triad is

unbalanced and creates tension, leading to pressure for change.

1. Two positive ties and one negative tie (++−): A likes B, B likes C, but A
dislikes C. This is unstable. A might try to convince B to dislike C, or A
might start to dislike B, or A might change their mind and start to like
C.

2. Three negative ties (−−−): A dislikes B, B dislikes C, and A dislikes C.

This is also unstable. For instance, if A and B are enemies, and B and C
are enemies, it creates tension if A and C are also enemies. It's often
"the enemy of my enemy is my friend."

• Example: Let's consider three individuals: Alice (A), Bob (B), and Carol (C).

o Balanced Triad (+++):

▪ Alice likes Bob (+)

▪ Bob likes Carol (+)

▪ Alice likes Carol (+)

 ▪ Result: This is a stable, harmonious group. (Product of signs: + * + * +
= +)

o Balanced Triad (+--):

▪ Alice likes Bob (+)

▪ Bob dislikes Carol (-)

▪ Alice dislikes Carol (-)

▪ Result: This is also stable. Alice and Bob are united in their dislike for
Carol. (Product of signs: + * - * - = +)

o Unbalanced Triad (++-):

▪ Alice likes Bob (+)

▪ Bob likes Carol (+)

▪ Alice dislikes Carol (-)

▪ Result: This triad is unstable. Alice might feel uncomfortable that her
friend Bob likes someone she dislikes. Pressure will exist for one of the
relationships to change to restore balance (e.g., Alice starts to like
Carol, or Bob starts to dislike Carol, or Alice starts to dislike Bob).
(Product of signs: + * + * - = -)

o Unbalanced Triad (---):

▪ Alice dislikes Bob (-)

▪ Bob dislikes Carol (-)

▪ Alice dislikes Carol (-)

▪ Result: This is also unstable ("the enemy of my enemy is my friend"

principle is violated). Alice and Bob are enemies, and Bob and Carol
are enemies. It's illogical for Alice and Carol to also be enemies; they
might become friends to collectively oppose Bob, or one of the
relationships might change. (Product of signs: - * - * - = -)

(b) Differentiate between Structural equivalence and Regular equivalence. (7.5 marks)

This question requires a good understanding of both concepts and their nuances.

Structural Equivalence
 • Definition: Two actors (nodes) are structurally equivalent if they have exactly the
same pattern of ties to and from all other actors in the network. This means they are
connected to precisely the same set of third parties.

• Strictness: This is a very strong and often too strict definition for real-world social
networks. It's rare to find perfectly structurally equivalent actors, especially in large
networks, due to the high likelihood of minor differences in connections.

• Implication: If two actors are structurally equivalent, they occupy identical positions
in the network. They are interchangeable from a purely structural perspective. They
would be exposed to the same information, constraints, and opportunities.

• Example:

o Imagine a small company with two receptionists, "R1" and "R2".

o If R1 reports to "Manager A" and communicates with "Client X", "Client Y",
and "Vendor Z".

o If R2 also reports to "Manager A" and communicates with "Client X", "Client
Y", and "Vendor Z".

o In this hypothetical scenario, R1 and R2 are structurally equivalent because

their connections to all other nodes are identical.

• Use Case: Often used in smaller, highly formalized networks or as an ideal type for
theoretical analysis.

Regular Equivalence

• Definition: Two actors (nodes) are regularly equivalent if they have the same pattern
of ties to and from some set of actors, who are themselves regularly equivalent. In
simpler terms, they have similar types of relationships with similar types of others,
even if they aren't connected to the exact same individuals.

• Flexibility: This is a much more flexible and realistic concept for social network
analysis than structural equivalence. It acknowledges that people can have similar
roles even if their specific contacts differ.

• Implication: Actors who are regularly equivalent occupy the same "role" or "status"
within the network. They perform similar functions or have similar responsibilities,
experiencing similar network constraints and opportunities. They might not interact
with the same exact people, but the types of people they interact with are the same.

• Example (continuing from above):

o Consider the same company. R1 reports to Manager A and communicates

with Client X, Client Y.
 o R2 reports to Manager B and communicates with Client P, Client Q.

o R1 and R2 are not structurally equivalent (they don't connect to the exact
same people).

o However, if Manager A and Manager B are regularly equivalent (i.e., they

perform similar managerial roles and connect to similar types of
employees/clients), and Client X, Y, P, Q are all regularly equivalent (i.e., they
are all "clients"), then R1 and R2 could be considered regularly equivalent.
They both fill the "receptionist" role: reporting to a manager and
communicating with clients.

• Use Case: Widely applied in studying social roles, organizational structures, and
identifying functional positions in large, complex networks.

Key Differentiators in a Table Format:

Feature Structural Equivalence Regular Equivalence

Similar types of ties to and from similar

Identical ties to and from the
Criterion types of other actors (who are
exact same other actors.
themselves regularly equivalent).

Very strict; rarely found in More relaxed and practical; commonly

Strictness
real, large networks. found in real networks.

Identical
Focus Similar roles/functions/status.
positions/connectivity.

Actors are truly Actors are interchangeable in terms of

Interchangeability
interchangeable structurally. their role.

Two doctors who both interact with

Two clones with identical
Example Scenario patients, nurses, and other doctors, but
friend lists.
not the same patients, nurses, etc.

Formalized systems, small Organizational roles, social classes,

Applications
networks, theoretical models. functional positions in large networks.

Export to Sheets

same for 4

Let's tackle Question 4 from the image, keeping the exam point of view and marks in mind (5
marks for (a), 7.5 for (b)).
Question 4

(a) What is data mining? How can it be used in social network analysis? (5 marks)

• What is Data Mining?

o Definition: Data mining is the process of discovering patterns, insights,

anomalies, and valuable information from large datasets using a combination
of techniques from statistics, machine learning, and database systems. It's
about extracting meaningful knowledge that was previously unknown and
potentially actionable.

o Goal: The primary goal of data mining is to turn raw data into useful
information that can be used for decision-making, prediction, or
understanding complex phenomena. It often involves steps like data cleaning,
integration, selection, transformation, mining itself (applying algorithms), and
pattern evaluation.

• How can it be used in Social Network Analysis (SNA)?

o Data mining techniques are extensively used in SNA to extract meaningful

patterns and insights from the massive and complex datasets that represent
social networks.

o 1. Relationship Extraction & Network Construction: Data mining algorithms

can be used to identify and extract relationships between entities from
unstructured or semi-structured data sources (e.g., text from emails, social
media posts, web pages). For instance, natural language processing (NLP)
techniques can identify mentions or interactions to build a network of
communication or influence.

o 2. Community Detection: Algorithms like k-means, hierarchical clustering, or

spectral clustering (common in data mining) can be applied to network data
to automatically identify communities, groups, or clusters of nodes that are
more densely connected to each other than to nodes outside the group. This
helps in understanding subgroups within a larger network.

o 3. Link Prediction: Data mining models (e.g., classification algorithms like

decision trees, support vector machines, or matrix factorization) can be
trained on existing network data to predict the likelihood of new connections
forming between nodes in the future. This is useful for friend
recommendations, predicting collaborations, or identifying potential
relationships in criminal networks.

o 4. Influence and Centrality Analysis: While SNA has its own centrality
measures, data mining algorithms can augment this by identifying influential
 nodes or information spread patterns. For example, anomaly detection
techniques can flag unusually influential nodes or unusual patterns of
information flow.

o 5. User Profiling and Behavioral Analysis: By combining network structure

with user attributes, data mining can create rich profiles of users, segment
them based on their network behavior, and predict their preferences,
interests, or future actions within the social network.

o 6. Anomaly Detection (e.g., Fraud, Spam): Data mining techniques can

identify unusual patterns in network connections that might indicate
malicious activity, spam bots, or fraudulent transactions in social networks.

(b) Explain the different types of centrality measures with help of example. (7.5 marks)

Centrality measures are fundamental in SNA as they help identify the most "important" or
"influential" nodes within a network based on different definitions of importance.

• 1. Degree Centrality:

o Definition: The simplest measure. It's the number of direct connections

(edges) a node has.

o Interpretation: In an undirected network, it measures popularity or activity

(how many direct contacts a node has). In a directed network, we distinguish
between:

▪ In-degree: Number of incoming connections (e.g., how many people

follow you, how many times a paper is cited). Often indicates prestige
or popularity.

▪ Out-degree: Number of outgoing connections (e.g., how many people

you follow, how many papers you cite). Often indicates gregariousness
or activity.

o Example: In a friendship network, a person with a high degree centrality has

many friends. If it's a "following" network, a high in-degree means many
followers (popular), and a high out-degree means they follow many others
(active).

• 2. Betweenness Centrality:

o Definition: Measures the extent to which a node lies on the shortest paths
between other nodes in the network. A node with high betweenness
centrality acts as a "bridge" or "broker" for information flow.
 o Interpretation: These nodes are critical for connecting different parts of the
network. Removing them can disrupt communication or connectivity
significantly. They have control over information flow.

o Example: In an organization, a person who is the only link between two

different departments or teams would have high betweenness centrality.
Information (or resources) from one department to the other would have to
pass through them.

• 3. Closeness Centrality:

o Definition: Measures how close a node is to all other nodes in the network.
It's calculated as the inverse of the sum of the shortest path lengths from that
node to all other nodes.

o Interpretation: Nodes with high closeness centrality can reach other nodes in
the network more quickly. They are well-positioned to spread information
efficiently or access information quickly.

o Example: In a rumor spreading scenario, a person with high closeness

centrality could spread a rumor throughout the network very quickly because
they are "close" to everyone else. A node that is physically or socially distant
from many others would have low closeness.

• 4. Eigenvector Centrality:

o Definition: Measures a node's influence based on the influence of its

neighbors. A node's eigenvector centrality is proportional to the sum of the
eigenvector centralities of its neighbors. In simple terms, connecting to highly
connected/influential nodes increases your own centrality.

o Interpretation: This measure indicates how well-connected a node is to other

well-connected nodes. It's a measure of being connected to the "right" people
or important sources.

o Example: In a network of academic citations, a paper that cites many other

highly cited papers (and is itself highly cited) would have a high eigenvector
centrality. On social media, being followed by many influential accounts (even
if your raw follower count isn't the highest) would boost your eigenvector
centrality.

These four are the most commonly discussed centrality measures, each offering a different
perspective on a node's importance within a network.

Sources

also of 5
Okay, here's a breakdown of Question 5 from the image, keeping the exam perspective and
marks in mind (5 marks for (a), 7.5 for (b)).

Question 5

(a) What is information diffusion? Explain different types of it. (5 marks)

• Definition: Information diffusion is the process by which information, ideas,

innovations, or behaviors spread through a social network or population over time. It
examines how and why information is transmitted from person to person, and how it
is adopted (or rejected) by individuals.

• Types of Information Diffusion:

1. Simple Contagion:

▪ Analogous to the spread of a simple disease. Exposure to the

information from even a single infected/informed source is sufficient
for an individual to adopt it.

▪ Examples: Basic factual information, widely known news, simple

behaviors.

2. Complex Contagion:

▪ Requires reinforcement from multiple sources before an individual

adopts the information or behavior. The more people you know who
have adopted it, the more likely you are to adopt it.

▪ Examples: Adoption of new technologies, complex behaviors, changes

in social norms, joining a social movement.

3. Broadcasting:

▪ Information is disseminated from a central source to a large audience.

The audience members don't necessarily interact with each other.

▪ Examples: Traditional mass media (TV, radio), public announcements.

4. Cascades:

▪ A specific type of diffusion where a small initial set of adopters

triggers a chain reaction, leading to widespread adoption.

▪ Examples: Viral marketing campaigns, online trends, sudden adoption

of a new product.

5. Hybrid Models:
 ▪ Real-world information diffusion often involves a combination of these
types. For example, a new product launch might involve broadcasting
(through ads) combined with complex contagion (word-of-mouth).

(b) What are the different types of social network analysis? (7.5 marks)

• Whole Network Analysis:

o Focuses on the entire network and the relationships among all actors within a
defined boundary.

o Aims to understand the overall structure, patterns, and dynamics of the

complete network.

o Examples: Analyzing the entire friendship network within a school, studying

the collaboration network of all scientists in a specific field, mapping the
entire communication network of a small company.

• Ego Network Analysis:

o Focuses on a single "ego" (a focal individual) and their direct connections

("alters").

o Examines the structure and characteristics of the ego's personal network.

o Examples: Studying the support network of a cancer patient, analyzing the

professional connections of a CEO, mapping the social circle of a teenager.

• Semantic Network Analysis:

o Focuses on the relationships between concepts, words, or ideas rather than

people.

o Represents knowledge as a network of interconnected concepts.

o Examples: Analyzing the relationships between keywords in a text corpus,

mapping the semantic structure of a website, studying the connections
between symptoms and diseases.

• Two-Mode (Affiliation) Network Analysis:

o Examines the relationships between two different types of entities (e.g.,

people and events, actors and movies).

o Analyzes how these two sets of entities are connected.

 o Examples: Studying which people attend which events, analyzing which actors
appear in which movies, examining which board members sit on which
company boards.

• Dynamic Network Analysis:

o Focuses on how networks evolve and change over time.

o Examines the formation, dissolution, and modification of relationships over

time.

o Examples: Studying how a social movement grows, analyzing how

communication patterns change within a team over a project's lifespan,
modeling the spread of a disease over time.

• Online Social Network Analysis:

o Specifically focuses on social networks that exist online (e.g., Facebook,

Twitter, LinkedIn).

o Examines the unique characteristics and dynamics of online interactions.

o Examples: Analyzing information diffusion on Twitter, studying community

formation on Facebook, mapping professional connections on LinkedIn.

• Spatial Network Analysis:

o Considers the geographical location of nodes and the spatial aspects of

network connections.

o Examines how geographical proximity influences relationships.

o Examples: Studying the spread of a disease based on geographical location,

analyzing trade networks between cities, mapping transportation networks.

• Weighted Network Analysis:

o Focuses on the strength or intensity of relationships between nodes.

o Examines how the varying weights of connections influence network

dynamics.

o Examples: Analyzing communication networks where the frequency of

interaction is considered, studying trust networks where the level of trust is
quantified, examining economic networks where the volume of trade is
measured.

for 6
Alright, let's break down Question 6 from the image, keeping the exam perspective and
marks in mind (5 marks for (a), 7.5 for (b)).

Question 6

(a) What is Graph theory? Explain the different ways to represent a graph. (5 marks)

• What is Graph Theory?

o Definition: Graph theory is a branch of mathematics concerned with the

study of graphs, which are mathematical structures used to model pairwise
relations between objects. A graph typically consists of a set of vertices (or
nodes) and a set of edges (or links) that connect pairs of vertices.

o Purpose: It provides a powerful framework for modeling and analyzing

various kinds of relationships and structures in diverse fields, including
computer science, social sciences, biology, and engineering. It allows us to
understand the connections, flow, and properties within complex systems.

• Different Ways to Represent a Graph:

1. Adjacency Matrix:

▪ Description: An N×N matrix (where N is the number of vertices) where

Aij=1 if there is an edge from vertex i to vertex j, and Aij=0 otherwise.
For weighted graphs, Aij could represent the weight of the edge. For
undirected graphs, the matrix is symmetric (Aij=Aji).

▪ Pros: Easy to check for the existence of an edge between two vertices
(O(1) time), easy to add/remove edges, good for dense graphs (many
edges).

▪ Cons: Wastes space for sparse graphs (many zeros), adding/removing

vertices is complex.

▪ Example:

▪ ABCD

▪ A|0110

▪ B|1001

▪ C|1000

▪ D|0100

(For an undirected graph where A-B, A-C, B-D are edges)

2. Adjacency List:

▪ Description: An array of lists (or vectors) where the i-th element of

the array stores a list of all vertices adjacent to vertex i.

▪ Pros: Space-efficient for sparse graphs (stores only existing edges),

easy to add/remove edges, efficient for traversing neighbors of a
vertex.

▪ Cons: Checking for the existence of an edge can be O(degree) in the

worst case, but typically faster on average.

▪ Example: (Same graph as above)

▪ A: [B, C]

▪ B: [A, D]

▪ C: [A]

▪ D: [B]

3. Edge List:

▪ Description: A list of all edges in the graph. Each edge is represented

as a pair (for undirected) or an ordered pair (for directed) of vertices.
For weighted graphs, a triplet (u, v, weight).

▪ Pros: Very simple to implement and understand, memory-efficient for

sparse graphs.

▪ Cons: Not efficient for operations that require finding all neighbors of
a vertex or checking for edge existence, as it requires iterating through
the entire list.

▪ Example: (Same graph as above)

▪ [(A, B), (A, C), (B, D)]

(b) Explain the different types of graphs. (7.5 marks)

Graphs can be classified based on various characteristics of their edges and vertices.

• 1. Undirected Graphs:

o Description: Edges have no direction. If an edge connects vertex A to vertex

B, it means A is related to B, and B is related to A in the same way. The
relationship is reciprocal.
 o Example: Friendship networks (if A is friends with B, B is friends with A),
family relationships (siblings).

• 2. Directed Graphs (Digraphs):

o Description: Edges have a direction, indicated by an arrow. If an edge goes

from A to B, it means A has a relationship with B, but B does not necessarily
have the same relationship with A.

o Example: "Follows" relationships on social media (you can follow someone

who doesn't follow you back), citation networks (paper A cites paper B, but B
doesn't necessarily cite A), organizational hierarchies (manager supervises
employee).

• 3. Weighted Graphs:

o Description: Edges have a numerical value (weight or cost) associated with

them, representing the strength, distance, time, or capacity of the
relationship.

o Example: Road networks (weights could be distance or travel time between

cities), communication networks (weights could be frequency of interaction),
social networks (strength of friendship).

• 4. Unweighted Graphs:

o Description: Edges do not have any associated numerical value. The existence
of an edge simply indicates a connection.

o Example: Simple friendship networks where you just know if two people are
friends or not.

• 5. Simple Graphs:

o Description: No loops (an edge connecting a vertex to itself) and no multiple

edges (more than one edge between the same pair of vertices).

o Example: Most basic social network models.

• 6. Multi-graphs (or Pseudo-graphs):

o Description: Allow multiple edges between the same pair of vertices (e.g.,
two people might communicate via email and phone, represented by two
edges) and/or loops (e.g., a person communicating with themselves).

• 7. Connected Graphs:

o Description: In an undirected graph, it means there is a path between every

pair of vertices. In a directed graph, it can be strongly connected (a path from
 every vertex to every other vertex) or weakly connected (the underlying
undirected graph is connected).

o Example: A single, continuous road network where you can drive from any
city to any other city.

• 8. Disconnected Graphs:

o Description: Consist of two or more separate "components" or "islands"

where there is no path between vertices in different components.

o Example: Two separate groups of friends who have no mutual acquaintances.

• 9. Complete Graphs (K_n):

o Description: Every distinct pair of vertices is connected by a unique edge.

o Example: A very tightly-knit group where everyone is directly connected to

everyone else (a clique).

• 10. Bipartite Graphs:

o Description: The vertices can be divided into two disjoint and independent
sets, such that every edge connects a vertex in one set to one in the other
set. No edges within the same set.

o Example: A graph showing students and the courses they are enrolled in
(edges connect students to courses, not students to students or courses to
courses).

• 11. Trees:

o Description: A connected, undirected graph with no cycles. It has N vertices

and N-1 edges.

o Example: Family trees (hierarchical relationships), organizational charts.

Sources

for 8

It seems like Question 7 is missing from the image you provided, as we've jumped directly
from Question 6 to a request for Question 8.

Could you please provide Question 7 first, if you'd like an answer for it?

If you are certain that you want to skip Question 7 and proceed with Question 8, please
provide the text of Question 8 (parts a and b), and I'll be happy to answer it for you!

for 7
Ah, my apologies! You are absolutely right. I misread the numbering on the image. Let's
tackle Question 7 now, keeping the exam perspective and marks in mind (5 marks for (a), 7.5
for (b)).

Question 7

(a) What is data processing and analysis? (5 marks)

• Data Processing:

o Definition: Data processing is the act of converting raw data into a more
usable and understandable form. It involves a series of operations performed
on data to make it organized, structured, and ready for analysis. It's the
preparation phase before insights can be extracted.

o Key Stages:

1. Collection: Gathering data from various sources.

2. Cleaning/Validation: Identifying and correcting errors,

inconsistencies, duplicates, and missing values in the data. This is
crucial for data quality.

3. Transformation: Converting data into a suitable format for analysis.

This might involve normalization, aggregation, feature engineering, or
converting data types.

4. Storage: Storing the processed data in a secure and accessible manner

(e.g., databases, data warehouses).

• Data Analysis:

o Definition: Data analysis is the process of inspecting, cleansing, transforming,

and modeling data with the goal of discovering useful information, informing
conclusions, and supporting decision-making. It applies statistical and
computational techniques to extract insights from processed data.

o Key Aspects:

1. Exploratory Data Analysis (EDA): Summarizing main characteristics of

data, often with visual methods, to understand patterns, detect
outliers, and test hypotheses.

2. Statistical Analysis: Applying statistical methods to test hypotheses,

identify relationships, and make inferences (e.g., correlation,
regression, hypothesis testing).
 3. Predictive Analysis: Using statistical and machine learning models to
forecast future outcomes or probabilities based on historical data.

4. Prescriptive Analysis: Recommending actions based on insights

derived from data.

o Relationship: Data processing is a prerequisite for data analysis. You cannot

perform meaningful analysis on raw, unorganized, or dirty data. Processing
makes the data amenable to analysis.

(b) Explain the different ways to discover community structures in networks. (7.5 marks)

Discovering community structures (or clusters/modularity) in networks is a crucial task in

social network analysis. A "community" is typically defined as a group of nodes that are
more densely connected to each other than to nodes outside the group.

Here are different ways (algorithms/approaches) to discover community structures:

• 1. Modularity Maximization:

o Concept: This is one of the most popular approaches. It aims to partition a

network into communities such that the number of edges within
communities is significantly higher than expected by chance, and the number
of edges between communities is low. The "modularity" (Q) score quantifies
the strength of a division of a network into communities.

o Algorithms:

▪ Louvain Method: A widely used and fast algorithm that iteratively

optimizes modularity. It works in two phases: first, it optimizes
modularity locally for each node, then it aggregates communities into
super-nodes and repeats the process.

▪ Newman's Algorithm (Fast Greedy Algorithm): Iteratively joins

communities in a way that maximizes modularity at each step.

o Pros: Conceptually clear, provides a quantitative measure (modularity), often

finds good community structures.

o Cons: Resolution limit (may not find small communities), can be

computationally intensive for very large networks.

• 2. Edge Betweenness Centrality (Girvan-Newman Algorithm):

o Concept: This method is based on the idea that edges connecting different
communities (inter-community edges) will have high betweenness centrality
(i.e., many shortest paths between nodes pass through them). By iteratively
 removing edges with the highest betweenness centrality, the network
eventually breaks into communities.

o Process:

1. Calculate betweenness centrality for all edges.

2. Remove the edge with the highest betweenness centrality.

3. Recalculate betweenness centrality for all remaining edges.

4. Repeat until no edges are left or desired number of communities is found. The best
partition is typically chosen when modularity is maximized.

o Pros: Can reveal hierarchical community structure as edges are removed.

o Cons: Computationally very expensive, especially for large networks, due to

repeated recalculations of betweenness centrality.

• 3. Label Propagation Algorithms (LPA):

o Concept: A fast and simple algorithm that works by "propagating" labels

(community assignments) through the network. Each node adopts the label
that the majority of its neighbors have.

o Process:

1. Initialize each node with a unique label (its own community).

2. Iteratively, each node updates its label to the label that appears most frequently
among its neighbors.

3. The process continues until no node changes its label, indicating a stable community
structure.

o Pros: Extremely fast and scalable to very large networks.

o Cons: Can sometimes produce slightly different results on different runs (non-
deterministic if ties are broken arbitrarily), may not always find the optimal
community structure.

• 4. Hierarchical Clustering (e.g., Agglomerative Clustering):

o Concept: Builds a hierarchy of clusters. Agglomerative methods start with

each node as its own cluster and then iteratively merge the closest clusters
until a single cluster remains or a stopping criterion is met. The "closeness"
can be defined by various distance metrics in the network.
 o Process: Create a dendrogram (tree-like diagram) that shows the sequence of
merges. Communities are identified by cutting the dendrogram at a certain
height.

o Pros: Provides a visual hierarchy of community structures, allowing for

analysis at different levels of granularity.

o Cons: Computationally intensive for very large networks, choice of distance

metric and cut-off point can influence results.

• 5. Spectral Clustering:

o Concept: Uses the eigenvalues and eigenvectors of a graph's Laplacian matrix

to embed the nodes into a lower-dimensional space where they can be easily
clustered using traditional clustering algorithms (e.g., k-means).

o Pros: Can discover non-linearly separable clusters, effective for detecting

communities that are not necessarily well-separated by simple cuts.

o Cons: Computationally more complex, particularly for very large graphs,

requires choosing the number of clusters (k) beforehand in some
implementations.

Each method has its strengths and weaknesses, and the choice often depends on the
network's characteristics, the size of the data, and the specific goals of the analysis.

Sources

for 8

Okay, let's address Question 8 from the image, keeping the exam perspective and marks in
mind (5 marks for (a), 7.5 for (b)).

Question 8

(a) What is data warehouse? Explain its characteristics. (5 marks)

• What is a Data Warehouse?

o Definition: A data warehouse is a large, centralized repository that stores

current and historical data from multiple disparate operational systems within
an organization. It's designed specifically for reporting and data analysis, not
for day-to-day transaction processing. Its primary purpose is to support
business intelligence (BI), analytics, and decision-making by providing a
unified and consistent view of the organization's data over time.
 o Purpose: To consolidate data from various sources into a single,
comprehensive, and clean dataset, enabling users to perform complex
queries, generate reports, and conduct analytical studies without impacting
the performance of operational systems.

• Characteristics of a Data Warehouse:

1. Subject-Oriented:

▪ Data is organized around major subjects of the enterprise (e.g.,

customers, products, sales, suppliers), rather than around individual
applications or operational processes. This makes it easier for analysts
to find relevant data for specific business areas.

▪ Example: All data related to "customers" (from sales, service,

marketing) is integrated and stored together.

2. Integrated:

▪ Data from various heterogeneous source systems is integrated,

transformed, and loaded into a consistent format. Inconsistencies,
duplicates, and different data representations are resolved during the
ETL (Extract, Transform, Load) process to ensure data quality and
uniformity.

▪ Example: Customer IDs from different operational systems are

harmonized into a single, consistent customer ID in the data
warehouse.

3. Time-Variant:

▪ Data in the warehouse is always associated with a specific point in

time. It stores historical data, allowing for trend analysis, comparisons
over time, and tracking changes. It retains information that might be
deleted from operational systems.

▪ Example: You can analyze sales trends for a specific product over the
past five years, or compare customer behavior this quarter versus last
quarter.

4. Non-Volatile:

▪ Once data is loaded into the data warehouse, it is generally

permanent and does not change. Data is added incrementally and is
not updated or deleted in real-time. This ensures that historical
records are preserved and consistent for analysis.
 ▪ Example: A sales record from five years ago will remain unchanged,
allowing for reliable historical analysis.

(b) What are the different types of data warehouse models? Explain with help of diagram.
(7.5 marks)

Data warehouse models define the logical structure for storing and organizing data within a
data warehouse. The most common models are the Star Schema, Snowflake Schema, and
Fact Constellation Schema.

• 1. Star Schema:

o Description: This is the simplest and most widely used data warehouse
model. It consists of a central "fact table" surrounded by several "dimension
tables," resembling a star.

▪ Fact Table: Contains quantitative measures (facts) and foreign keys to

dimension tables. Facts are typically numerical and additive (e.g., sales
amount, quantity sold, profit).

▪ Dimension Tables: Contain descriptive attributes related to the facts.

They provide context to the facts (e.g., customer details, product
descriptions, time periods, location information).

o Characteristics:

▪ Denormalized: Dimension tables are typically denormalized to

improve query performance, meaning redundant data might exist
(e.g., product name and category in the same table).

▪ Simple Joins: Few joins are needed to retrieve data, making queries
fast.

o Diagram:

o [Time Dimension]

o | (Date, Year, Month, Day_of_Week)

o |

o [Product Dimension] --<-- [Fact Table: Sales] -->-- [Customer Dimension]

o (Product_ID, Name, Category) | (Time_ID, Product_ID, Customer_ID,

Store_ID, Sales_Amount, Quantity) | (Customer_ID, Name, Address, Age)

o |
 o |

o [Store Dimension]

o (Store_ID, Name, City, Region)

(Arrows indicate foreign key relationships from fact to dimension tables)

• 2. Snowflake Schema:

o Description: An extension of the star schema where dimension tables are

further normalized into multiple related tables. This creates a "snowflake" like
structure, as the branches of the star schema are themselves broken down
into sub-branches.

o Characteristics:

▪ Normalized Dimensions: Dimension tables are normalized, reducing

data redundancy.

▪ More Complex Joins: Queries might require more joins to access data,
potentially affecting performance compared to the star schema.

▪ Better for Complex Dimensions: Suitable when dimensions have

many attributes and clear hierarchical relationships that benefit from
normalization.

o Diagram:

o [Time Dimension]

o | (Date, Year, Month, Day_of_Week)

o |

o [Product Dimension] --------->-- [Fact Table: Sales] -->-- [Customer Dimension]

o (Product_ID, Name) | (Time_ID, Prod_ID, Cust_ID, Store_ID,

Sales_Amount) | (Customer_ID, Name)

o | | |

o V | V

o [Category Dim] [Store Dimension] [Address Dim]

o (Category_ID, Name) (Store_ID, Name) (Address_ID,

Street, City)

o | | |

o V V V
 o [Subcategory Dim] [Region Dim] [State Dim]

o (Subcat_ID, Name) (Region_ID, Name) (State_ID,

Name)

(Note: Prod_ID links Product Dim to Fact, Category_ID links Product Dim to Category Dim,
etc.)

• 3. Fact Constellation Schema (or Galaxy Schema):

o Description: This model involves multiple fact tables sharing one or more
common dimension tables. It's essentially a collection of interconnected star
schemas or snowflake schemas.

o Characteristics:

▪ Complex Business Processes: Used when there are multiple related

business processes that need to be analyzed together, sharing
common dimensions.

▪ Shared Dimensions: Dimensions like "Time," "Product," or

"Customer" can be shared across different fact tables, providing a
consistent view.

▪ Increased Complexity: More complex to design and manage due to

multiple fact tables and shared dimensions.

o Diagram:

o [Time Dimension]

o |

o [Product Dimension] --->-- [Fact Table: Sales] -->-- [Customer Dimension]

o |

o |

o V

o [Store Dimension]

o |

o |

o [Product Dimension] --->-- [Fact Table: Inventory] -->-- [Warehouse

Dimension]

o |
 o |

o V

o [Time Dimension]

(In this example, "Product Dimension" and "Time Dimension" are shared between "Sales"
and "Inventory" fact tables.)

Each model has its own advantages and disadvantages in terms of query performance,
storage space, and ease of maintenance, and the choice depends on the specific analytical
requirements and the nature of the data.

Please provide the content of Question 9 (parts a and b) from your image. I will then be able
to answer it for you.

yes

You mentioned "yes" but did not provide the content of Question 9. To help you, I need to
see the question.

Please type out or describe Question 9 (parts a and b) from your image, and I will then
provide the answer according to exam point of view for B.Tech.

the ques saya what is analysis of text? what are the four features of textual analysis what are
the steps in text analysis

It appears Question 9 has been combined into a single, comprehensive query about
"Analysis of Text." Let's address this multi-part question from an exam perspective, aiming
for a detailed answer suitable for a B.Tech exam.

Question 9: What is analysis of text? What are the four features of textual analysis? What
are the steps in text analysis?

This question combines three related aspects. I'll structure the answer accordingly.

(a) What is Analysis of Text? (Approx. 2-3 marks for definition)

• Definition: Text analysis, also known as text mining or text analytics, is the process of
extracting meaningful, quantitative, and qualitative insights from unstructured
textual data. It involves using computational and statistical methods to identify
patterns, themes, sentiments, relationships, and other valuable information that
would be difficult or impossible to discern manually from large volumes of text.

• Purpose: The primary goal is to transform raw, human-readable text into structured
data that can be further processed, analyzed, and used for decision-making,
 understanding trends, gaining customer insights, or conducting research across
various domains like social sciences, business intelligence, and natural language
processing.

(b) What are the four features of Textual Analysis? (Approx. 4-5 marks)

When discussing "features of textual analysis," we often refer to key characteristics of the
data itself or the types of insights it can provide. Here are four important features:

1. Unstructured Data: The most defining feature. Textual data (e.g., emails, social
media posts, articles, reviews) is typically unstructured, meaning it doesn't fit into a
pre-defined data model or schema like traditional databases. Text analysis is
designed specifically to make sense of this raw, free-form data.

2. Contextual Richness: Text carries immense contextual information, including

nuances, emotions, opinions, and relationships that are embedded within the
language. Unlike numerical data, text analysis aims to preserve and extract this rich
context to understand the meaning behind the words, not just their frequency.

3. Volume and Variety: Modern textual data often comes in enormous volumes (Big
Data) and diverse formats (e.g., short tweets, long research papers, dialogues,
reports). Textual analysis techniques must be scalable to handle this volume and
versatile enough to process different linguistic styles and structures.

4. Implicit Information/Pattern Discovery: A core feature is its ability to uncover

implicit patterns, hidden relationships, and latent topics that are not explicitly stated.
For example, it can identify emerging trends in customer complaints, detect
fraudulent patterns in documents, or reveal underlying sentiment towards a brand,
even if no direct "sentiment score" is present in the original text.

(c) What are the steps in Text Analysis? (Approx. 6-7 marks)

Text analysis is an iterative process that generally follows these steps:

1. Data Collection:

o Description: Gathering the raw textual data from various sources.

o Examples: Scraping web pages, downloading social media feeds, extracting

data from internal documents, accessing public datasets, or conducting
surveys with open-ended questions.

2. Text Pre-processing (Data Cleaning and Preparation):

o Description: This is a crucial and often time-consuming step to transform raw

text into a clean and standardized format suitable for analysis. It prepares the
text for algorithms.
 o Sub-steps:

▪ Tokenization: Breaking down text into individual words or phrases

(tokens).

▪ Lowercasing: Converting all text to lowercase to treat "The" and "the"

as the same.

▪ Stop Word Removal: Eliminating common words (e.g., "a", "an",

"the", "is", "are") that carry little semantic meaning for analysis.

▪ Stemming/Lemmatization: Reducing words to their root form (e.g.,

"running," "runs," "ran" to "run") to consolidate variations.

▪ Punctuation/Special Character Removal: Cleaning noise from the

text.

▪ Handling Missing Data/Noise: Addressing irrelevant or malformed

text.

3. Feature Engineering/Representation:

o Description: Converting the pre-processed text into a numerical or structured

format that can be understood by machine learning algorithms.

o Techniques:

▪ Bag-of-Words (BoW): Represents a document as a collection of its

words, disregarding grammar and word order, but keeping track of
word frequencies.

▪ TF-IDF (Term Frequency-Inverse Document Frequency): Assigns

weights to words based on their frequency in a document relative to
their frequency across all documents, giving more importance to
unique terms.

▪ Word Embeddings (e.g., Word2Vec, GloVe): Represent words as

dense vectors in a continuous vector space, capturing semantic
relationships and context.

4. Analysis/Modeling:

o Description: Applying various computational and statistical techniques

(algorithms) to the structured textual data to extract insights.

o Common Techniques:

▪ Sentiment Analysis: Determining the emotional tone (positive,

negative, neutral) of a text.
 ▪ Topic Modeling: Discovering abstract "topics" that occur in a
collection of documents (e.g., Latent Dirichlet Allocation - LDA).

▪ Named Entity Recognition (NER): Identifying and classifying named

entities (e.g., people, organizations, locations, dates) in text.

▪ Text Classification: Categorizing text into predefined classes (e.g.,

spam detection, customer complaint type).

▪ Clustering: Grouping similar documents or sentences together.

▪ Relationship Extraction: Identifying semantic relationships between

entities.

5. Evaluation and Interpretation:

o Description: Assessing the quality and validity of the results obtained from
the analysis and drawing meaningful conclusions.

o Activities:

▪ Evaluating model performance (e.g., accuracy, precision, recall for

classification).

▪ Visualizing results (e.g., word clouds, topic trends, network graphs of

relationships).

▪ Domain expert review to ensure insights are contextually relevant and

actionable.

▪ Refining the process based on initial findings.

This comprehensive approach allows for robust extraction of knowledge from vast amounts
of textual data.