Social Network Analysis

Open Elective IV

Presented by : Ms. Drashti S.

 Syllabus
Lecture No. Contents Duration(Hr)

1 Introduction to social network mining 1

2 Community detection 1

3 Shingling algorithm 1

4 Social Networks as Graphs 1

5 Random graph models 1

6 Ranking algorithms 1
Graph and Matrices,Basic measures for individuals and
7 1
networks
 Lecture 1
- Key Definitions :
a. Big Data
b. Community Detection
c. Data Mining
d. Social media mining
- Introduction to Social Network Mining
- Use of Social Media Data
 Key Definitions
1.Big data: Big data is data that contains greater variety, arriving in increasing volumes
and with more velocity.
2.Community detection: These techniques are useful for social media algorithms to
discover people with common interests and keep them tightly connected.
3.Data mining: It is the process of uncovering patterns and other valuable information
from large data sets.
4.Social media mining: It is the process of obtaining big data from user-generated
content on social media sites and mobile apps.
 Social Network Mining - The Plot
• In social media data mining, information from social media platforms is aggregated and analyzed to
identify patterns and trends.
• Social media platforms use algorithms to track and organize page views, likes, clicks and other user
interactions, and then apply the collective data to inform research and marketing processes, such as
targeted advertising.
• Third-party data-mining companies can also purchase or scrape data from numerous social media sites
(Facebook, Twitter, Instagram, etc.) along with forums, blogs, news sites and other public pages
where users leave comments or interact with specific social groups.
 Social Network Mining
• The combined unstructured data — profiles, posts, images, connections — can then be used to identify
behavioral patterns.
• Data miners sell datasets to companies, including airlines, hotels, retailers, manufacturers and political
groups, that are looking to personalize customer experiences and improve service satisfaction.
• While companies have collected user data for years, advancements in artificial intelligence have
enabled social media data mining to become increasingly sophisticated.
 Social Network Mining
• Machine learning and advanced cognitive technologies can not only analyze vast quantities of data,
they can also interpret jargon, acronyms and tone to capture context and provide a better gauge of user
sentiment.

• Because of this, companies are increasingly using social intelligence to inform product, marketing and
customer service decision-making.
• However, businesses must walk a fine line between catering to specific audiences and coming off as
overly intrusive.

• Laws around data mining are still evolving, and businesses must largely police themselves to remain
ethical. To protect user privacy, social media platforms have policies that govern what kinds of
information third parties can use.
 Use of Social Media Data in Business
• Businesses can use social media data mining in many ways as follows :

1. Advertising
• Targeted advertising on social media platforms is skyrocketing as companies figure out
better ways to identify and address specific audiences.
• Advertising executives can also use data-mining techniques to determine which messages
are most effective among certain demographic groups or the best time of day to run a
specific ad on a specific digital platform.
 Use of Social Media Data in Business
2. Influencer Marketing

• Social media data mining can be helpful in identifying influencers or users who have strong follower
bases and engagement rates on certain social platforms.
• Businesses will use influencer marketing to help bring attention to products and services.
• An influencer may be a high-profile company executive, a celebrity, a blogger or an external product
reviewer who can drive hits and clicks via a previously untapped sales channel. Careful analysis of
social data can help companies to find the right influencer to promote their offerings.
 Use of Social Media Data in Business
3. Market Research

• Companies use social media data mining to find out more about customer tastes, preferences and
biases.
• For example, an organization may want to examine the demographic characteristics of emerging client
groups or determine public sentiment toward a certain brand or logo — or even a politician or
religious group.

• Companies also use social media data to gather intelligence on specific geographies, competitors or
potential partners.
 Use of Social Media Data in Business
4. Sales Enablement

• In addition to gathering information on proprietary products, a business may gather social intelligence
on prospective customers’ or partners’ offerings to leverage in a sales pitch. A computer component
manufacturer may investigate what complaints are surfacing on a PC manufacturer’s goods, for
instance, in hopes of helping the client to improve brand sentiment.
5. Predictive Analytics

• Sophisticated algorithms and machine learning techniques can help with predictive modeling
strategies, enabling companies to anticipate future customer trends. Social media data is also helpful to
the medical community when tracking and predicting the course of disease outbreaks.
 Lecture 1 : Takeaway
Q.1 What Social Network Mining?
Q.2 How does Social Media become useful to Business?
 Lecture 2
- Community Detection: Definition
- Network Modularity
- Giant Component
- Community Detection: Detection
- Community Detection: Techniques
 Community Detection : Definition
• The people with common interests living in a
particular area broadlyor a group of people
with a common characteristic or interest living
together within a larger society.
• Community detection in a social network is
identifying sets of nodes in such a way that the
connections of nodes within a set are more than
their connection to other network nodes.
 Community Detection : Need
• When analyzing different networks, it may be important to discover
communities inside them.
• Community detection techniques are useful for social media algorithms
to discover people with common interests and keep them tightly
connected.
• Community detection can be used in machine learning to detect groups
with similar properties and extract groups for various reasons.
• For example, this technique can be used to discover manipulative groups
inside a social network or a stock market.
Modularity
Giant Component : Largest component in graph
Network Modularity
• Modularity is a measure of the
structure of networks or graphs which
measures the strength of division of a
network into modules (also called
groups, clusters or communities).
 Community Detection : Techniques
1. Louvain Detection : (detected in 2008)
• The algorithm starts with a weighted network of N nodes.
• In the first phase, the algorithm assigns a different community to each node of the
network. Then for each node, it considered the neighbours and evaluate the gain of
modularity by removing the particular node from the current community and placing
in the neighbour’s community. The node will be placed in the neighbour’s
community if the gain is positive and maximized. The node will remain in the same
community if there is no positive gain. This process is applied repeatedly and for all
nodes until no further improvement is there. The first phase of the Louvain algorithm
then stops.
• In the second phase, the algorithm builds a new network considering communities
found in the first phase as nodes.
 Community Detection : Techniques
2. Leiden Detection:
• According to them, Louvain has a tendency to discover very weekly connected
communities. Therefore, they have proposed the much faster Leiden algorithm which
guarantees that communities are well connected with a Refinement phase.
• In the refinement phase, the algorithm tries to identify refined partitions from the
partitions proposed by the first phase. Communities proposed by the first phase may
further split into multiple partitions in the second phase. This randomness allows
discovering the partition space more broadly.
• Also in the first phase, Leiden follows a different approach to the Louvain. Instead of
visiting all the nodes in the network after the first visit to all nodes has been
completed, Leiden only visits those nodes whose neighbourhood has changed.
 Community Detection : Techniques
2. Leiden Detection:
 Lecture 2 : Takeaway
Q.1 What is Community?
Q.2 What is the need for Community?
Q.3 Explain the techniques for Community Detection.