SOCIAL MEDIA ANALYTICS
PRESENTED BY:
TEAM: C10
ATHUL JOHN
[Link]
1
VAISHNAVI.S
AKHILA.S
HARESH.G
VENKAT RAMAN
ABHIRAM
2 INTRODUCTION TO SOCAL MEDIA
ANALYTICS
Used as a channel to communicate with stake holders.
Allows to way communication
Social media is monitored ,data are collected and analyzed for marketing
research.
The social media analytics leverages available social data to generate
marketing insights.
Provides guidance for new product innovation and development.
FIRM ACTVITY
3
SOCIAL MEDIA DATA
SOCIAL MEDIA
CONSUMER ACTIVITY
ANALYTICS
MARKETING INSIGHTS
OTHER DATA SET
4 CHARACTERTICS OF SOCIAL MEDIA
DATA
Volume of Data
Computational challenge.
Network structure
Users position may affect the outcome
Tags applied by the user
May reveal contributor's favorability
Privacy
Open platforms
Challenges in social Media
5
Analytics
6 Data Preparation
Collecting the relevant data
Convert unstructured data into structured data
Analysing the data
[Link] soft ware
[Link]
Representatives of Social Media Data
7
Very little control over the data
Consumer decide
[Link] they will participate in the conversation
[Link] of their contribution
[Link] which platforms they will contribute.
Ying, Feinberg, and Wedel research on how previous product reviews effect the users
decssion.
Volume and variance of previously contributed opinions affects these two processes.
contributors may discuss any aspect of a brand, product, or service.
8 Data Integration
Integrating data from different sources.
Firm ability to relate social media activity to other sources.
Firm ability to relate these data is limited.
Diversification in customer needs.
Privacy concerns of the consumer
BEYOND SOCIAL MEDIA
9
LISTENING
10
BEYOND SOCIAL MEDIA LISTENING
Social media listening is the process of identifying and assessing
what is being said about a company, individual, product or brand on
the Internet.
It combines the capabilities of social media listening with analytics.
They further identify areas in which they believe there are
opportunities for innovation, both in research and implementation.
11 BEYOND SOCIAL MEDIA LISTENING
(cont’d)..
Brand Management: Supplementing Traditional Marketing Research
Research has demonstrated the potential to use social media
analytics as a means of understanding perceptions of the brand and
how such perceptions may shift over time.
Traditional marketing research tools are well established and
provide the gold standards.
12 BEYOND SOCIAL MEDIA LISTENING
(cont’d)..
Crisis Management: Forecasting Social Media Conversations
It’s a report of what has already transpired rather than a projection of
what is expected to occur in the future.
Customer Relationship Mangement: Social profiling and Trageting
Develops customer relationships and monitor past media behavior in
order to target service interventions.
13 BEYOND SOCIAL MEDIA LISTENING
(cont’d)..
Product Development and Innovation
It provides insights about consumer preferences which are
particularly valuable for product innovation and development
processes.
Evaluating Other Marketing Actions
Combines data with other marketing activities to evaluate their
impact on metrics of interest including awareness, traffic, and
ultimately purchasing behavior.
Application of Social Media
14
Analytics
SOCIAL MEDIA ANALYTICS IN
15
HOSPITALITY AND TOURISM
Comparative examination on three online review platforms:
i. Trip Advisor
ii. Expedia
iii. Yelp
The goal to provide a basis for understanding
the methodological challenges and for identifying opportunities for the development of
social media analytics in hospitality and tourism.
RESEARCH DESIGN AND
ANALYTICAL FRAMEWORK
16
Data collection and Data
17
Analysis
18 Data Collection
Took place in late 2015 on all searchable hotel properties in
Site Data
Manhattan and NYC through Volume
Trip 439K
Web crawlers were used to mimic a user’s access Advisor
Expedia 481K
Several types of data were collected
Yelp 31k
Expedia had highest reviews and Trip Advisor had high
reviews per hotel
Data Analysis
19
Data Analysis
Examine
Data Pre Development of relationship
Processing Key metrics between key
measures
Review
Review
Review Topics helpfulness
Sentiment
of identification score
Identification
development
20 Data Analysis – Data Pre Processing
English reviews were collected
Tokenization is a form of lexical analysis whereby a stream of text is broken up into
words, phrases, or other meaningful elements called tokens
Each review was broken up into vector of unigram-based tokens using function called
Regex Tokenizer
Review length was computed for each review
Readability score % was measured
21
Data Analysis - Development of Key Metrics
Review Topics identification
Latent Dirichlet Allocation used to review text to discover the main topics
Output score was assigned to represent review containing a specific topic
Review Sentiment identification
Text mining procedure to discover emotive content
Naïve Bayes method was used
Each review was assigned a sentiment score between 0 and 1 with 0 and 1 representing the
two extremes of sentiment
Review helpfulness score development
Key indicator of review quality
centroid-based summarization approach used to compare the three platforms
Result of computing assigns a value of 0-1 depending on the helpfulness of each review
22
Data Analysis- Examine relationship between
key measures
Two linear regression models used to compare the three platforms
Both regression analyses were run in the JMP statistical software by SAS
Series of centralization operations were conducted
Ensured the variance inflation factor (VIF) scores for all predictors in the
regression models were below 10, which effectively removed the
multicollinearity problem
23
Findings
24 FINDINGS
When the star-rating system of Expedia was applied to Trip advisor &
Yelp , discrepancies were caused.
Discrepancies might be caused due to duplicates in downloaded data sets.
Avg. reviews were higher in Trip advisor & Yelp . Also helpful response to
reviews were higher in them.
87.5% of reviews in Expedia were of less than 50 words . People
expressed more in Trip advisor & Yelp.
25 FINDINGS : (contd.)
People expressed their dissatisfaction more in Yelp . This is reasonable as Yelp is a
platform for local business.
Most of the reviews was about basic service followed by core product.
Reviews about basic service was more prominent in Yelp and it had less reviews
about landmark & attraction. Value related reviews were more prominent in
Expedia.
Reviews with positive sentiments was more in Expedia and Trip advisor. Thus
Yelp had more polarizing customers.
26 FINDINGS : (contd.)
Branded hotels had better rating than other hotels. But the difference was small to a factor
of 0.036.
People find review length to be helpful. If the review length is short they mostly look at
other review characteristics.
27
DISCUSSIONS AND IMPLICATIONS
28 DISCUSSIONS
Due to the lack of understanding in social media-related studies,
three major review platforms such as TripAdvisor, Expedia, and
Yelp are used.
In terms supply of hotel product, there appears to be huge
discrepancies between these platforms.
In terms review data, TripAdvisor and Expedia are comparable to
each other, while Yelp is smaller.
In terms of overall sentiment, TripAdvisor and Expedia are similar
while Yelp is quite unique.
29 IMPLICATIONS FOR RESEARCH
First, directly drawing data from online websites must, indeed, consider the
inherent traits and biases in social media data.
Second, there is a growing interest in understanding the communicative
effects of online reviews within and outside hospitality and tourism.
Third, this study reveals the intricacies between some of the variables
associated with online reviews.
Finally, this study also sheds light on the structure of guest experience as
revealed in online reviews and its connection with satisfaction.
30 IMPLICATIONS FOR PRACTICE
This study clearly shows that not all review websites are created equal;
they vary considerably in quality and focus.
Hospitality and tourism businesses must make wise .decisions when
choosing these channels to engage with their existing and prospective
customers.
Therefore, businesses should develop effective means to encourage
customers to share their experiences in rich and constructive ways
31 CONCLUSION AND FUTURE RESEARCH
Information technology creates new structures and dynamics in the market; therefore, it
is important for us to gain a solid understanding.
In this study we showed that online review data drawn from three dominant platforms.
The analytical framework applied in this study can be expanded to include more
theoretically meaningful variables.
Therefore, this study will help develop the methodological and theoretical foundations
for social media analytics in hospitality and tourism and beyond.
32
