Social Media Analytics

Semester: 08 (R-19)
Course Code: CSDC8023
(Credits: 03)
Dr. Jayant Nandwalkar
 Module 5

Social Information
Filtering
 Curriculum
5 Social Information Filtering
Social Information Filtering – Social
Sharing & Filtering, Automated
Recommendation Systems. Traditional Vs
Social Recommendation Systems
Understanding Social Media and Business
Alignment, Social Media KPI, Formulating
a Social Media Strategy, Managing Social
Media Risks.
 Social Information Filtering
Among the ample data, some of it can be more
useful than the rest.
To accomplish the same, sorting of data is required.
Social sharing & social filtering is one of way to
find the useful information among the links, news,
videos & photos posted.
Information filtering uses semi-automated or
computerized methods to remove redundant or
unnecessary information from an information stream
before presenting it to human user.
 Social Information Filtering
Its main goal is to manage the information overload
& increase the semantic signal-to-noise ratio.
E.g. spam in emails
Recommender systems & content discovery
platforms are active information filtering system, it
attempts users with information that interests them
(movie, television music, books, news, websites)
rather than removing information flowing to the user.
Filtering approaches: Content-based approach
& Collaborative Filtering approach.
Recommendation uses either one approach or
combined approach
 Social Information Filtering
How do peoples handles this filtering ?
Most items & information are removed from
stream because they are inaccessible or invisible to
the user.
E.g. Newspaper editors. (from bulk news to
shortcut news)
E.g. Book stores decides which books to carry.
All above mentioned things are through the
technology (software) is nothing but the handling of
information filtering.
 Social Sharing & Filtering
Social sharing websites, like Digg, Slashdot &
reddit are designed for people to share interesting
content.
Community then votes items up or down & the
most interesting links are highlighted.
Large no. of peoples to help complete the task is a
kind of crowed sourcing. From this crowed of online
peoples each contributes work by sharing or voting
on content & the aggregate results are valuable
contribution.
Majorly social news sites shares the information
When users interest is limited more to sharing with
his or her fiends, Facebook or Tweeter may be used
where the links can be shared.
 Social Sharing & Filtering
Some times including Facebook, are also tracking
users’ behavior to share links to content they are
viewing, without user taking any active action to share
Frictions Sharing is a stream coined by face book
founder Mark Zukenberg to describe this trend, which is
present on Facebook as well as many new sites & web
portals.
This type of sharing is most often seen in applications
called social readers.
When user activates a social reader, either through a
social network or a new source, the application lists all
the articles that the users clicks on. The user takes no
action to share these clicks; instead, the application
shares everything to user does.
 Social Sharing & Filtering
This raise the privacy concern for many users who do
not want all their activity shared.
Aggregating behavior that may show the “top-viewed”
articles without information about who viewed them is
one way around the privacy issues.
Tips For Increasing Social Shares:
1. Find out which social networks your target audience
use the most. If the customer base is primarily business
& corporations, pay attention to LinkedIn. Facebook is
more social, casual relationships and better suited for
online shopping customers. Tweeter tends to walk the
line between casual & formal.
2. Create meaningful content. The worst thing one can
do is become a spammer on SM networks. Sending link
s & posting to boards should be done carefully.
 Automated Recommendation Systems:
Recommender systems are major parts of e-commerce sites
and social media sites.
Even if recommender system is not familiar, but shortly all
internet users will be familiar.
Recommender systems are the features of website that
suggest items a user might like.
Amazon.com uses one to suggest other items a customer
might want to buy.
Netflix uses it to suggests movies that a subscriber might
want to check out.
Pandora uses it to automatically generate internet music
channels that match a user’s taste.
 Automated Recommendation Systems:
All of these personalized suggestions based on a user’s previous
activity come from recommender systems.
Recommender system is a machine learning technique that
suggests relevant content to user based on their characteristics &
past behaviors within the system. These solutions are gaining the
popularity in e-commerce, advertising, entertainment and more
recently, training apps & platforms.
Common examples of recommendation systems includes apps like
spotify & Netflix, where the system learns and recommends songs
and movies based on the likelihood that a user will like content or
apps based on browsing behaviors, purchase history & intensions.
For example, Amazon recommends products. Relevance suggests a
purchase.
 Automated Recommendation Systems:

Powerful recommendation systems (Applications

of recommendation system) in industries such as:

1.Retail & e-commerce

2.Entertainment subscription services content
recommendation.
3. Travel sector & hyper personalization of products
and services.
4. News website & social networks Banking services
for new products.
Advantages of Recommendation Systems:
 Advantages of Recommendation Systems:
Recommendation system can do many things for the
business.
1. Drive Traffic: A recommendation engine can bring
traffic to your site. It accomplishes this with customized e-
mail messages & targeted blasts.
2. Provide Relevant Material: By analyzing the customer’s
present site use and his previous browsing history, a
recommendation engine can deliver appropriate product
suggestions as he stores. The data is gathered in real time
so the software can respond as his shopping habits change.
3. Engage Customers: Consumers end up being more
engaged in the website when individualized item
recommendations are made. They are able to dive even
more deeply into the product line without needing to carry
out search after search.
 Advantages of Recommendation Systems:
4. Transfer Shoppers to Clients: Converting buyers into
consumers takes a unique touch. Individualized
communications from a recommendation engine reveal
your customer that he is valued as an individual. In turn
this engenders his loyalty.
5. Increase Average Order value: Average order values
generally go up when a recommendation engine in uses to
show tailored alternatives. Advanced metrics and
reporting can definitively reveal the efficiency of project.
6. Boost Number of Items per Order: In addition to the
average order value rising, the number of products per
order likewise typically increases when a
recommendation engine is used. When a customer is
revealed options that fulfill his interest, he is mostly
likely to add choices to his purchase.
 Advantages of Recommendation Systems:
• 7. Control Retailing & Inventory Rules: A
recommendation engine can add your very own marketing
& inventory control directives to the customer’s profile to
feature items that are promotionally prices, on clearance
or overstocked. It offers you’re the versatility to regulate
exactly what items are highlighted by the recommendation
system.
8. Lower work & Overhead: The volume of data
required to create an individual shopping experience for
each customer is typically far too huge to be handled
manually. Utilizing an engine automates this process,
reducing the workload of IT staff & spending plan.
 Advantages of Recommendation Systems:
• 9. Provide Reports: Providing guides is an integral part of a
personalization system. Providing the client precise & up to the
minute reporting permits him to make solid choices about his
website & the direction of project.
• 10. Offer Recommendations & Direction: An experienced
carrier can provide suggestions on ways to utilize the data
gathered & reported to customer. Acting as a partner &
consultant, the supplier needs to have the expertise to assist
direct the e commerce site to prosperous future.
How to Build a Recommendation Systems:
 Types of Recommendation System
 1. Collaborative Filtering: This recommendation system is based on
inputs or actions from other people (rather than only the user for whom
a recommendation system is being made).
By User Similarity: This strategy involves creating user groups by
comparing users’ activities & providing recommendations that are
popular among other members of the group. It is useful on sites with a
storng but versatile audience to quickly provide recommendations for a
user on which little information is available.
By Association: This is a specific type of the one mentioned above,
otherwise known as “Users who looked at X also looked at Y”.
Implementing this type of recommendation system is a matter of
looking at purchasing sequences or purchasing groups & showing
similar content. This strategy is useful for capturing recommendations
related to naturally complementary content as well as a certain point in
the life of the user.
 Example: Movies on Netflix
Pros: Continuous Learning
required
Cons: Rating feedback
is required &
New items and
users face a cold start

V1: Viewer 1, V2: Viewer 2, 5*: Five Stars

V1 saw movie Hangover & given 5*, V2 saw movie Hangover & given 5*
 V1 saw movie Project X & given 5*, V2 saw movie Project X & given 5*
Now V2 saw movie The stone Man Murders & given 5*
Recommender system of Netflix then will list “The stone Man
Murders” for V1 because both have same types of choices with
same ratings means liking of both is same. Accordingly the
expected rating from V1 by the Netflix for the same movie is 5*
 Content Based Recommendation
• Definition: Recommended items to customer X similar to
previous items rated highly by X i.e. uses item features to
recommend new items that are similar to what the user has
liked in the past.
• Example:

• Likes
• Item Profile: User profile: Red, Circle, Rectangle.
• Next Recommendations : Match

• Unmatch

Here recommendation system shown Red

• Pros: No community requirement
Hexagon & Green Circle because in item
• Cons: Needs of content description profile red, circle & rectangle are there &
• New user face cold start these are irrespective of combination that
is what these recommendations are.
 Traditional Vs Social Recommendation Systems
• Example:
Name of viewer Star Wars Jaws Wizards OZ God Father 2001 Movie
ALICE 5 4 3 3 1
BOB 3 5 2 5 1
CHUNK 4 3 2 2 2
• Table shows the ratings by three users for 5 different movies with 1-5
scale.
• One method to compute the similarity: Average difference between the
ratings of these movies.
• BOB has average difference is 1.2 [(5&3), (4&5), (3 &2), (3&5),
(1&1)] with ALICE, while CHUNK has an average difference as
1.0[(5&4), (4&3), (3&2), (3&2), (1&2)]. A more common measure of
similarity is the correlation between the ratings.
 Traditional Vs Social Recommendation Systems
• CHUNKS ratings are 1 point lower than ALICE for every movie
except 2001, where it is 1 higher. His rating track very closely with
ALICE. On other side BOBs rating does not follow any pattern.
• Correlation Coefficient: A statistic that measures how well
aligned 2 sets of values are.
• Now assume ALICE wants to know how much she might like
movie Vertigo which she has never seen. Both BOB & CHUCK
have seen it. BOB rated it as 3 & CHUCK rated it as 5.
• What would be good recommendation to ALICE for how much she
will like it?
• Show the average rating for the movie, which is a 4. (it is
irrespective of CHUCK is more similar to ALICE than BOB is but
the average of both 3 & 5).
 Traditional Vs Social Recommendation Systems
A simple example of collaborative filter will use the
correlation numbers to compute a weighted average. BOB
& CHUCK’s rating will be multiplied by their correlation
with ALICE. & then total will be divided by sum of the
weights.
The weighted average comes out higher than the simple
average. That is because CHUCH gets more weight & since
he is more similar to ALICE, his higher rating of movie is
given more consideration. Thus the recommended rating of
Vertigo for ALICE is 4.5 stars. This is recommender
system rating.
The ratings produced by recommender system can be
used directly to indicate to user how much they like a
particular item or it can be used to sort or filter the items.
Understanding Social Media & Business Alignment
or
What is business Alignment ?
• It is simply not enough to have a social media analytics tool ready
to mine data. Analytics should be strategically aligned to support
existing business goals. Without a well-crafted and aligned social
media strategy, your business will struggle to get the desired
outcomes from analytics.
• Key points to understanding this: concepts of alignment, social
media alignment matrices, role of chief information officer (CIO)
in facilitating the alignment & steps needed to formulate social
media strategy.
 Business Alignment
 Definition: Business alignment is the process in
organization used to ensure that system & elements
works together to support long term goals.
 Aligning social media objective & goals with the
objectives of the organization should be the starting point
of any SM analytics initiative.
 Business alignment I show professionals align elements
of business so that they can work together to achieve
business goals.
 Business alignment helps in improving i) streamlining of
business ii) employees collaboration iii) efficiency of
business iv) better & enhanced productivity v) business
outcomes vi) Profit vii) retaining of customers. Viii)
resources
 Types of Business Alignment

1) Business IT alignment: It enables IT to align strategy

with business leadership team. This alignment allows
IT department to leverage its technical expertise to
update computer system so that business objectives can
meet effectively or efficiently.
2) Alignment of Management System: It enables to
operate process & policies. It automates the tasks &
record data which are the fundamental part of business.
By connecting management system one can establish
standard operating procedures for each system & give
employees clear instructions in various situations. One
can focus on the development activities instead putting
efforts in managing & storing the data.
 Types of Business Alignment

3) Alignment of Resources: Allocation of resources such as

budget (proper fund management) . This alignment avoids
the wasting of money in unwanted specifics.
4) Alignment of Communication: It clarifies the
communication between employees & the company. This
alignment creates the open line communication so that
every one can come to know about important news.
5) Reward and Recognition: It includes how company
reward & recognize their employees. Motivation to
employees must be there by rewarding best performers in
organization. Rewards can be through salary,
commissions & these should be aligned with goals of
organization.
 Social Media KPI
Growth or the achievement of goals of business can be
determined by proper alignment of SM along with various
factors including the availability of technical, financial,
administrative & leadership resources & its potential to
achieve business goals.
Simplified SM analytics alignment matrix:
Y axis: Resource availability (financial, technical,
administrative & leadership resources for social media
analytics)
X axis: Impact of SM analytics alignment in terms of its
potential to achieve business goals (or potential to generate
economic value & return on investment). Depending on 2
variables (i.e. resource availability and its potential) the SM
analytics alignment with business goals can fall into 4
possible quadrants.
 Social Media Alignment Matrix

This matrix can be flexible because one can replace the variables of both
axis:
X axis: Sensitivity of the analytics ?(security, privacy or ethics).
Y axis: Criticality of SM analytics (extent to which the analytics is critical to
the business).
 Formulating a Social Media Strategy
The purpose of formulating SM strategy is to create rules
and procedures to align SM engagement with business goals.
Steps in formulating a Social Media Strategy
1.Get Hold of an Executive Champion: success is possible
only if the transformation is steered through strong
leadership: setting direction, building momentum & ensuring
the disciplined execution of an inspiring vision and ambitious
plans. SM executive champion should be the personality with
charisma & power to enforce social media plans. (Usually the
HOD or government chief information officer)
A champion should have several attributes including
power, vision, resources & willingness to sacrifice &
persistence to support the change.
 Formulating a Social Media Strategy
2. Build a Cross-Functional Team: Team includes senior
management members from all the department including IT
department. Team must be lead by CIO. CFT function: all the
stakeholders are independent to say & have their ownership to
social media analytics initiative.
3. Assess the organizational culture: Understanding organization's
vision, values, norms, systems, assumptions & beliefs about
social media is very critical. To fulfill the same Culture Builder
Toolkit is developed by corporate culture pros.
4. Review Your Current Social Media Presence: Organization
should monitor the social media floating.
5. Determine the objectives:
6. Aligning Social Media Goals with Business Goals:
7. Develop your content Strategy:
8. Platform Strategy:
9. Resource Considerations:
 Managing Social Media Risks
1. Risk Identification:
2. Risk Assessment:
3. Risk Mitigation
4. Risk Evaluation:
Thank You