2.

Social Network Structure, Measures &

Visualization
Part -II
Why do we need Network Measures?
Centrality
● Centrality is one of the core principles of network analysis. It measures how
“central” a node is in the network. This is used as an estimate of its
importance in the network.
● Depending on the application and point of view, what counts as “central”
may vary depending on the context.
● Four ways to compute Centrality-
○ Degree centrality,
○ Closeness centrality
○ Betweenness centrality
○ Eigenvector centrality.
Degree Centrality
● Is the degree of a node( i.e, the number of edges it has)
● The higher the degree, the more central the node is.
Centrality
Degree centrality is a
good measure of the
total connections a node
has, but will not
necessarily indicate the
importance of a node
in connecting others or
how central it is to the
main group.
Centrality
Find degree centrality of each node?
● Degree centrality of a
node refers to the
number of edges
attached to the node. In
order to know the
standardized score, you
need to divide each
score by n-1
Degree centrality of each node:
Find degree centrality of each node
Closeness centrality
● Closeness centrality indicates how close a node is to all other nodes in
the network.
● It is calculated as the average of the shortest path length from the
node to every other node in the network.
Closeness centrality
Since node D’s closeness centrality is (3+2+1+1+2+2+1)/7= 1.71 and node A’s is
3.43, node D is more central by this. It indicates nodes as more central if they are
closer to most of the nodes in the graph measure
In the case of
closeness centrality,
or average shortest
path length, lower
values indicate more
central nodes.
Find closeness centrality of node 8?
Betweenness centrality

● Measures how important a node is to the shortest paths through the

network.
● For Calculating Betweenness
○ We select a pair of nodes and find all the shortest paths between those nodes.
○ We compute the fraction of those shortest paths that include node N.
Eigenvector centrality

● Eigenvector centrality measures a node’s importance while giving

consideration to the importance of its neighbors.
● For example, a node with 300 relatively unpopular friends on Facebook
would have lower eigenvector centrality than someone with 300 very
popular friends (like Barak Obama).
● It is sometimes used to measure a node’s influence in the network.
● It is determined by performing a matrix calculation to determine what is
called the principal eigenvector using the adjacency matrix.
● Links from important nodes (as measured by degree centrality) are worth
more than links from unimportant nodes.
Describing networks

● Degree distribution is a way to describe the distribution of degrees

among the nodes in a network.
● To create a degree distribution, the degree of each node in the network
is calculated.
● The next step is to count how many nodes have each degree and
create a table or chart to display this information.
● A bar graph is a common way to display a degree distribution, with the x-
axis showing the degree and the y-axis indicating the number of nodes
with that degree.
Describing networks

● Degree distribution
Density
● Density describes how connected a network is.
● It is a statistic comparing the number of edges that exist in a network
to the number of edges that could possibly exist.
● Consider the following two networks, which both have the same number
of nodes.
○ Network (a) has very few edges while
○ Network (b) has numerous edges among the same number of nodes.
○ Network (b) has higher density.
Density
Density
Connectivity
● Density measures the percentage of possible edges in
a graph.
● Connectivity, also known as cohesion, measures how
those edges are distributed.
● Connectivity is a count of the minimum number of
nodes that would have to be removed before the graph
becomes disconnected.
Connectivity
The connectivity is
The connectivity is 2
1 because removing
node B, C, or D would
disconnect the graph.
Centralization
● Centralization measures the extent to which the ties of a given network
are concentrated on a single actor or group of actors.
● C(n) be the centrality of node n, using whatever centrality measure we
choose.
● n* is the most central node.
● Find the difference in centrality between n* and every other node in the
network, and add those up.
Centralization
● n* is the most central node.
● Find the difference in centrality between n* and every other node in the
network, and add those up.

divide this by the sum of the maximum possible differences

between n* and every other node.
Centralization

centralization =
Tie Strength
● Tie Strength is the measure of the strength of relationship between
the people.
● Strong Ties -rare , family members and close friends
○ Strong ties are more trusted and their information is more likely to be reliable.
● Weak Ties - includes acquaintances and more casual relationships.
○ Weak ties always connect to diverse groups of people with different
perspectives.
● Absent Ties - The people with whom we have no relations.
Granovetter’s four principals
Measuring tie strength-Granovetter’s four principals
● Time can include the amount of time people spend with each other, the
duration of their relationship (i.e., how long they have known each other), and
how frequently they see one another.
● Emotional intensity is indicated by the closeness of a relationship; close
friends or family members are likely to be strong ties, while more casual
friends and acquaintances would be weaker ties.
● Intimacy-relates to people sharing secrets or intimate
personal details with one another. The more of this information they
exchange, the closer their relationship is likely to be.
● Reciprocal Services are favors that people do for one another. They may be
personal (e.g., pet sitting or picking up someone’s dry cleaning), financial
(e.g., loaning money), professional (e.g., putting people in contact with one
another), or otherwise.
Measuring tie strength -other factors
● Structural features relate to the social network of the two people in question.
Those who have many mutual friends are likely to have stronger ties.
● Social Distance measures how different people’s social situations are. This
includes factors like age difference, race, education, and socioeconomic status.
People with strong ties tend to have similar social attributes.
● Emotional Support describes the communication between people that
validates their emotions, shows understanding of their problems, and tries to
alleviate stress.
Measuring tie strength
● Intimacy
● • Number of days since their last communication
● • Number of friends in common
● • Number of “intimate” words in their communications, as determined by
● software that automatically analyzes text
● Intensity
● • Number of words exchanged on one another’s walls
● • Depth of email threads in their inboxes (i.e., how many messages were sent
● back and forth in a conversation)
● Reciprocal services
● • Number of links shared on one another’s wall
● • Applications the users had in common (presumably because they could be
● working together within the application context)
● Social distance
● • Age difference
● • Difference in the number of educational degrees
● • Difference in the number of occupations
Network propagation
Tie strength and network structure

When the tie does not exist it is called as forbidden triad

Network propagation
● Network propagation is a phenomenon where things spread through a
network.
● How tie strength relates to the rate at which phenomena can spread
through a network?
Network propagation
● Granovetter -study about finding jobs.
● A person has more weak ties than strong ties, and because the weak
ties are connected to diverse social groups that the person would not
otherwise communicate with, the weak ties are more often sources of
new information, like job opportunities.
Tie strength and network propagation
● Weak ties are powerful in helping spread information farther through
a net-work, but this is not to say that strong ties are unimportant.
● Strong ties tend to be more trustworthy, reliable, and personal.
● A weak tie may be able to tell a person about many job opportunities,
but a strong tie will have a better idea of what jobs would be a good
fit.
● Strong ties are also more likely to do things to help one another (i.e.,
reciprocal services).
Trust
● A person trusts another if she is willing to take a risk based on her
expectation that the trusted person’s actions will lead to a positive
outcome.
● Trust is formed through a combination of factors such as successful
interactions, personal experiences, shared beliefs and values, and
the presence of guarantees and protections offered by institutions.
Trust
● 1. Calculation-based trust: This is a rational decision about whether to trust
someone, and where the costs and benefits of trusting are factored in.
● 2. Personal-based trust: This reflects a person’s propensity to trust,
developed over the course of their life.
● 3. Cognition-based trust: This describes the instant rapport and trust that
can develop between people who share similar backgrounds, beliefs, and
values. It often is based on first impressions.
● 4. Institution-based trust: This addresses how trust may form in the
presence of guarantees and protections offered by an institution.
Trust -Asymmetry
● Children must have almost absolute trust in their parents, while the parents may
have almost no trust in their children, particularly when they are very young.
● Context and time
● Someone may trust her friend to recommend a movie but not to repair her car.
● People tend to develop trust over time as their history of shared experiences
builds.
● Trust may also decrease.
● If someone has one dramatic failure, trust may disappear completely.
Measuring trust
1. Propensity to trust
● Refers to a person's inclination to trust others.
● Investment Game is a common way to measure a person's propensity to trust and the trustworthiness of others
2.Trust in others
● An individual's amount of trust in others varies based on their propensity to trust and their beliefs
● about the specific person they are deciding to trust.
● Decisions about trust in others deal with the risks a person is willing to take with a specific person.
○ 1. Trust with material possessions
○ 2. Belief about reliability
○ 3. Trust with secrets
○ 4. Trust regarding physical safety
Measuring trust
● Trust in social media is harder to judge compared to in-person interactions as
the information available about a person is limited and identity can be easily
forged.
● Before social media, users were mostly concerned with trusting websites,
especially e-commerce sites, and retailers tried to overcome this by creating
privacy policies, assuring transaction safety and security, and building
professional and secure websites.
● With the rise of social media and user interaction, trust issues shifted to include
both websites and other users.
● Reputation systems, such as the one used by eBay, were introduced to help
users make decisions about trust based on feedback from other users.
Measuring trust
● Social media users often share a lot of personal information, some of
which is private, and must trust both the website and other users to treat
their information with respect.
● Complex issues regarding trust in social media arise due to the
constantly changing factors of what users want to share, with whom
they want to share it, and what control they have over those
decisions.
● Reputation systems are not always present in social media applications
and understanding trust in social media is a complex and constantly
evolving topic.
Measuring trust
● The problem of inferring trust arises in social networks, where trust
between two people can be considered a weight on the edge
connecting them.
● The determination of trust is influenced by an individual's propensity
to trust and their assessment of the other person's trustworthiness.
● Surveys could be used to generate a trust rating, but most people will not
take the time to fill them out.
Network-based inference
Network-based inference
1. The social network contains nodes for each person and edges with trust ratings (1 to 10).
2. The network can have multiple paths between the Source and Sink.
3. Some paths may have high trust values while others may have low trust values.
4. The Source can use centrality measures to assess the trustworthiness of the Sink based on the information
in the social network.
5. One example of a centrality measure is the weighted harmonic centrality, which considers the weighted
distances between a node and all other nodes in the network.
6. Another approach is to use the PageRank algorithm, which assigns a score to each node in the network
based on the trust ratings of its neighbors.
7. The source would need to gather information about the trust ratings of all paths between the Source and the
Sink.
8. Based on this information, the source can calculate centrality measures or PageRank scores to make a
reasonably informed estimate of the Sink's trustworthiness.
Similarity-based trust inference
● People in social networks do not always supply trust ratings.
● Research shows that people who trust each other tend to be similar in some way.
● Similarity is important, but other factors such as major disagreements or
disagreements on important issues can impact trust.
● Trust can be estimated using similarities in item ratings, - When people have
provided ratings of items (e.g., they have rated movies),the similarities in their
ratings can be used to estimate trust.
● Estimating trust based on item ratings is an area of active research and refinement.
● Researchers consider various types of information such as ratings, profile
information, patterns of social media use, history of interactions, etc.
Network Visualization
● Network visualization, also known as graph visualization or link analysis, is
the process of visually presenting networks of connected entities as links
and nodes.
● Nodes represent data points, and links represent the connections
between them.
● Good visualization supports user in better understanding the data they
are seeing
● The goal is to take advantages of human’s natural abilities to see patterns,
anomalies, relationship and features in visual data
● Visualization provide an overview of complex data
● From there, people can identify features of interest, refocus attention on
those features and explore more
● Visualizations are a qualities way to begin understanding data. From
these quantitative experiments or analysis can follow to explain any
insights
● Graph visualization apply all these lesson to looking at the structure of
networks.
Graph Layout
● Layout algorithm that position the nodes and edges in different ways
when visualizing a network
● “good” layout depends on what the analyst wants to find, what to type of
network is being viewed and what its features
● General guidelines
○ Every node is visible
○ For every node you can count its degree.
○ For every link you can follow it from source to destination
○ Clusters and outliers are identifiable
● Random Layout
○ When loading data into visualization tool, the nodes are placed randomly.
○ And it often does not provide much insight into the structure of the network.
○ In random layout, identifying high degree in a node may be possible but the cluster and
other patters are not clear.
Circular Layout
● Place all nodes in a circle and then add edges between them.
● Some circular layouts place nodes closer to one another when they are
more closely connected.
● A circular layout places nodes in structured position and then adds edges
between connected pairs.
Grid layout
● where all graph nodes are placed on a 2-dimensional squared grid.
Force-directed layout
● a type of network graph layout that uses a physics-based simulation to
position the nodes and edges of the graph.
● Nodes and edges are treated as a physical system, and a simulation of
that system is applied to determine a final layout.
● For example, nodes may be treated as objects, and edges may be treated
as springs that apply equal force.
● The nodes are randomly laid out, connected by springs for edges, and
then a simulation of how the springs would physically behave determines
the final position of nodes and edges.
Other layout
Visualizing network features
● Edges, weights, node properties, labels and clusters can also be
visualized.
● Labels
○ More difficult attributes to show in network, both on edges and on nodes
● Size, shape and color
○ Keeping color as an indicator of degree, node size can be used to indicate clustering
coefficient
Scale Issues
● Visualization is very useful for analyzing networks of this size or smaller.
When networks become much larger, the quality of the visualization
diminishes.
Density
● Density can also be a problem for visualization, even if the number of
nodes is small.
Graph simplification
● Graph simplification techniques include grouping clusters of nodes into a
single node and representing the edges between clusters as a single edge,
representing structural patterns as representative shapes, or showing
only part of the graph at a time.
● Tree Structure:
○ The nodes and edges to the right of the first level are hidden but are summarized with
triangles. The size, color, and angle of the triangles indicate the depth, number of nodes,
and width of the summarized structure.
Social network analytics
● NETWORKS are the building blocks of social media and can carry useful business
insights.
● Social media network analytics thus deals with constructing, analyzing, and
understanding social media networks
● The purpose of network analysis is to
○ Understand overall network structure; for example, number of nodes, number of links, density,
clustering coefficient, and diameter.
○ Find influential nodes and their rankings; for example, degree, betweenness, and closeness
centralities.
○ Find important links and their rankings; for example, weight, betweenness, and centrality.
○ Find cohesive subgroups; for example, pinpointing communities within a network.
○ Investigate multiplexity; for example, analyzing comparisons between different link types, such
as friends vs. enemies.
Common Social Media Networks
● Friendship network
● Follow-Following Network
● FAN Network
● Group Network - people who share common interests and agendas.eg. Yahoo
Groups, Facebook Groups
● Professional Networks - Nodes in these networks are, for example,
people, brands, and organizations, and links are professional relations (such as
coworker, employee, or collaborator). An important feature of professional
networks is the endorsement feature, where people who know you can endorse
your skills and qualification.
Common Social Media Networks
● Dating Network
● Co authorship Network
● Hyperlink Network
● COLIKE -colike networks are formed when two or more people like the
same social media content.
● COCOMMENTER NETWORKS
● Cocommenter networks are formed when two or more people comment on socialmedia
content (e.g., a Facebook status update, blog post, or YouTube video).
● Nodes represent users, and link represents the cocommenting relationship.
TYPES OF NETWORKS
● BASED ON EXISTENCE -
○ Implicit (keyword co occurrence networks, co citation networks, cocommenter networks,
hyperlink networks)
○ Explicit network- (keyword cooccurrence networks, cocitation networks, cocommenter
networks, hyperlink networks)

● BASED ON DIRECTION
○ 1) directed networks
○ 2) undirected networks.
TYPES OF NETWORKS
● BASED ON MODE
○ Based on the composition of nodes, networks can be classified as
○ 1) One-mode network - A one-mode network is formed among a single set of nodes of the same
nature . A Facebook friendship network is an example of a one-mode network where nodes (people)
form network ties (friendships).

○ 2) Two-mode networks - Two-mode networks (also known as bipartite networks) are networks with
two sets of nodes of different classes.one set of nodes (circles) could be social media users and other
set of nodes (squares) could be participation in a series of events. Users are linked to the events they
attended.
○ 3) Multimode networks.
● BASED ON WEIGHTS
○ 1) weighted networks,
○ 2) unweighted networks.
TYPES OF NETWORKS
COMMON NETWORK TERMINOLOGIES
NODE-LEVEL PROPERTIES
● Degree Centrality
● Betweenness Centrality
● Eigenvector Centrality

NETWORK-LEVEL PROPERTIES
● Clustering Coefficient - The clustering coefficient of a network is the degree to which nodes in a network tend to cluster or
group together.
● Density -Density can be calculated as the number of links present in a network divided by the number
● of all possible links between pairs of nodes in a network
● Components -Components of a network are the isolated sub-networks that connect within, but
● are disconnected between, sub-networks.

● Diameter -The diameter of a network is the largest of all the calculated shortest path between any pair of nodes in a network
● Average Degree -The average degree centrality measures the average number of links among nodes in a
network.
NETWORK ANALYTICS TOOLS
● NodeXL: A free tool for social network analysis and visualization for networks on platforms like Facebook, Twitter,
and YouTube.

● UCINET: A social network analysis software for Windows operating system with a free 90-day trial.

● Pajek: A software application for analyzing and visualizing large networks, free for non-commercial use on
Windows operating systems.

● Netminer: A software application for large social network analysis and visualization, with a free 28-day trial.

● Flocker: A real-time Twitter analytics tool for retweets and mentions networks.

● Reach: An online platform for mapping hashtag networks and identifying the most influential accounts on Twitter.

● Mentionmapp: An online tool for investigating Twitter mentions networks.

NETWORK ANALYTICS TOOLS - NodeXL
Questions
● 1. What is a network?
● 2. What is the purpose of network analytics?
● 3. Briefly differentiate among social networks, social network sites, social
● networking, and social network analysis.
● 4. Briefly explain the different types of social media networks.
● 5. What is the difference between explicit and implicit networks?
● 6. What is the difference between one-mode, two-mode, and multimode
● networks?
● 7. Differentiate between weighted and unweighted networks.
● 8. Briefly define important node level properties, such as degree, betweenness,
● eigenvector centralities, and structural holes.
● 9. Briefly explain important network-level properties, such as clustering
● coefficient, density, diameter, average degree, and components.
● Explain the importance of Weak Ties and Strong ties
● What is tie strength and how to measure it?
● Explain the Granovetter’s four principles of measuring tie strength