1-MAYSSA

Table of contents
The presentation starts by introducing the problem we're addressing. We then look at what's
already been tried and critique those approaches. Next, we propose our solution and outline
what we aim to achieve. We'll explain the technology and architecture behind our solution.
After that, we'll show how our solution brings extra benefits, including a live demo. Finally,
we'll wrap up by discussing what comes next and our plans for the future
2-RCHID
Intoduction:
In the dynamic telecom industry, analyzing operators' data is crucial for informed decision-
making by operators, clients, and enterprises. Insights into market share, customer churn,
revenue trends, and network performance provide a comprehensive view, enabling operators
to optimize pricing, innovate services, and enhance customer satisfaction. Clients and
enterprises benefit by making informed choices about service providers, investments in
telecom infrastructure, and strategic partnerships. Robust data analysis supports all
stakeholders in navigating the complex telecom landscape and driving sustainable growth.
3-AMANI
Problem statement:
In the highly competitive telecom market of Tunisia, customers face significant challenges in
selecting the most suitable operator due to the fragmented presentation of offers across
multiple platforms. This situation is compounded by the difficulty telecom operators
experience in understanding their market position and customer feedback without a
centralized system to collect and analyze data from various sources. The primary question,
therefore, is:
How can a centralized and automated system be developed to streamline the collection,
comparison, and analysis of telecom offers and customer feedback from social media,
thereby enhancing customer decision-making and enabling telecom operators to
optimize their services and competitive strategies?
This problem encompasses the need for an efficient method to aggregate and analyze data
from disparate sources, providing both customers and operators with actionable insights.

4-BOUMNIJEL
Existing chalenges:
Fragmentation of Information: Offers from telecom operators are spread across multiple
websites, making comparison difficult for customers. Customers must manually navigate
each website to gather information, which can be lengthy and tedious. Lack of Social Media
Analysis: Although operators receive reactions and comments on their social media pages,
they often lack automated means to systematically collect and analyze this data. The absence
of analysis tools limits operators' ability to understand customer sentiments and needs.
Absence of Data Centralization: There is no centralized database where information on offers
and social interactions can be stored and analyzed together. Relevant data is often stored in
disparate locations, making analysis and extraction of insights more complex.

Conclusion
The lack of an integrated solution for collecting and analyzing telecom offers and customer
interactions on social media creates inefficiencies for both customers and operators.
Customers struggle to compare offers, and operators lack the tools to fully leverage customer
feedback. An automated and centralized solution for data collection and analysis is needed to
improve the customer experience and enable operators to better meet market demands.
Proposed solution
5-MAYSSA
Operator Data Insights is our sophisticated analytics platform, a revolutionary solution for
understanding Tunisia’s telecom landscape. By collecting data from social media and official
telecom operator websites, our platform delivers deep insights through advanced data
processing and visualization techniques
6-NADIM
 Automatic collection of targeted operator offers, customer feedback, posts, and reactions to
posts: This involves setting up automated systems to gather information such as special offers
from specific operators, feedback from customers, various social media posts, and reactions
to those posts.

 Data cleaning, processing, and data warehouse construction: Once the data is collected, it
undergoes cleaning and processing to ensure accuracy and consistency. This processed data is
then used to build a data warehouse, which serves as a centralized repository for organized
and structured data storage.

7-AMANI
 Sentiment analysis of feedback: Utilizing advanced techniques, feedback from customers
and social media posts are analyzed to determine sentiment. This helps in understanding
whether the feedback is positive, negative, or neutral, providing valuable insights into
customer satisfaction and public perception.

 Creation of dashboards: Based on the analyzed data, dashboards are created to visualize
key metrics and trends. These dashboards provide stakeholders with intuitive and easy-to-
understand summaries of the collected data, facilitating informed decision-making.

 Accessible website implementation: Developing and deploying a user-friendly website that

allows stakeholders to access relevant information and insights derived from the data. This
ensures that the data-driven insights are readily available and accessible to all authorized
users, promoting transparency and usability.

8-BOUMNIJEL

BUSINESS objectives

To ensure better quality of our application, we have complemented the functional

requirements with the following business requirements:
Security: Our application must be secure with an authentication system
The dashboards created must be well-represented, ergonomic, and meet the needs and
comfort for thz desision makers to achieve optimal performance.
Our application must provide better data integrity to ensure accurate results for effective
decision support.
Reliability: The system must ensure mission accomplishment and meet defined
specifications.
9-MAYSSA
Technical objectives
Authentication and Access: Ensure secure authentication mechanisms and controlled access
to the dashboards
Automated Data Collection: Develop a Python program that automates the process of
scraping telecom offers and social media interactions.
Data Centralization in a Data Warehouse (DWH): Consolidate all collected data in a
centralized Data Warehouse for efficient management and analysis.
Data Storage and Management: Implement a system to store the scraped data in JSON format
and securely upload it to Azure Blob Storage.
Preparation for Analysis: Ensure the collected data is structured and ready for sentiment
analysis in Power BI.

Technologies
10-NADIM
Comprehensive Data Collection
We harness the power of Python for robust web scraping, collecting extensive data from
social media platforms and the official websites of Tunisia's telecom operators. This data
includes user comments, reactions, likes, and post dates, offering a detailed view of customer
engagement and feedback.
11-AMANI
Secure Data Management
ODI integrates seamlessly with Snowflake, a premier data warehousing solution, to manage
your data securely and efficiently. Snowflake’s scalable architecture ensures that even the
largest datasets are handled with ease, maintaining data integrity and accessibility.
12-MAYSSA
Insightful Visualizations
From Snowflake, data is transferred to Power BI, where it is transformed into dynamic,
interactive dashboards. These visualizations provide comprehensive sentiment analysis and a
comparative overview of the various telecom operators' offerings, helping users make
informed decisions.
13-RCHID
Advanced Language Processing
Using Azure’s translation services, we ensure that comments and feedback in various
languages are accurately translated, enabling a more inclusive and thorough analysis. This
helps in understanding the sentiment and opinions of a diverse user base.
14-BOUMNIJEL
Sentiment Analysis
Our application leverages Azure AI services to perform comprehensive sentiment analysis.
By analyzing customer feedback and interactions, our platform provides deep insights into
customer sentiment. This ensures that users receive the most relevant information .
14-NADIM
Topic Extraction
Using machine learning algorithms, our application extracts the most discussed topics from
customer comments. By processing vast amounts of data, the platform identifies key themes
and trends, offering valuable insights into customer concerns and preferences.

16-MAYSSA
Architecture
Our architecture consists of multiple stages, beginning with data scraping, followed by
storage, processing, transformation, and finally analysis.
It all starts with data scraping from Facebook pages of top telecom operators using Python
and Selenium. We collect comments and feedback on both fixed and mobile categories. The
collected data is then pushed to Azure Blob Storage in JSON format. From there, the data is
loaded into Snowflake, a cloud data warehouse solution, through Snowpipe. This happens in
the first layer called the Data Landing Zone, or DLZ. Next, the data is transformed into the
Trusted Zone, or TRZ. Automated procedures run whenever new data arrives, ensuring
everything is up to date. Finally, the data reaches the Refined Zone, where Data Marts are
created for each operator. This refined data is then ready for analysis. We connect this refined
data to Power BI, where we perform sentiment analysis with the help of Azure AI's Text
Analytics service. This allows us to analyze feedback and translate sentiments accurately.
Each operator gets a customized dashboard, this ensures they see only their data and insights.

17-RCHID
Added value
Reputation Management
ODI (Operator Data Insights) enables operators to monitor customer sentiment towards their
services, pinpoint specific emotions from feedback, and engage with influential customers to
bolster brand reputation effectively.
Product Analysis
Operators leverage ODI to gather and analyze customer feedback on their service offerings.
This data enables them to enhance service features, discern customer preferences, and
benchmark performance against competitors.
Voice of Customer Analysis
By analyzing customer sentiments across diverse platforms, operators gain valuable insights
into customer preferences, needs, and expectations. This facilitates tailored service
adjustments that effectively meet customer demands.
18-BOUMNIJEL
Market Research Competitor Analysis
Operators utilize ODI to compare customer perceptions of their services with those of
competitors. This strategic analysis informs decision-making processes and optimizes service
positioning in the market landscape.
Customer Support-Feedback Analysis
ODI enhances customer support by evaluating feedback sentiments, automating responses to
common inquiries, and prioritizing timely resolution of customer issues.
Real-Time Customer Insights
Operators employ ODI to gather instantaneous insights from social media and customer
reviews. This capability enables swift identification of emerging issues, improvement of
service quality, and enhancement of overall customer satisfaction.
Product Development Optimization
Operators optimize their service offerings by analyzing customer sentiments and preferences
through ODI.
19-AMANI
Personalized Marketing Campaigns
ODI supports operators in crafting personalized marketing campaigns by deeply
understanding customer sentiments and preferences.
Predictive Analytics
Operators employ ODI for predictive analytics, forecasting customer behavior, market trends,
and service demand.
Risk Mitigation
By analyzing customer sentiment, operators can proactively mitigate risks, improve service
quality, and address potential issues promptly.
20-RCHID
Decision makers
Operator Users:
• High-Quality Services: Clients prioritize high-speed internet, mobile connectivity,
and attractive offers.
• Customer Support: Effective and responsive customer service is crucial. Operators
with a reputation for customer satisfaction are preferred.
Enterprises:
• Scalable Solutions: Enterprises need telecom services that grow with their business
and offer tailored solutions.
• Flexible Service Plans: Plans must adapt to changing demands and support large-
scale operations.
• Competitive Pricing: Cost-effective solutions are essential without compromising
quality for a better return on investment.
Other Operators:
• Service Offerings and Innovation: Studying competitors' portfolios and initiatives
helps identify trends and stay competitive.
• Network Performance Metrics: Assessing coverage, speed, and reliability guides
infrastructure investments and technological advancements.
21-AMANI
Conclusion
With ODI, you gain comprehensive insights into customer sentiment and feedback, allowing
you to effectively compare offerings from various telecom operators. This empowers you to
make informed decisions based on real-time data and provides telecom operators with
actionable feedback to enhance their services.
22-NADIM
Perspectives
Perspective
In the future, we aim to enhance our system by incorporating localization features and
expanding our scope to include telecom operators worldwide. This expansion will enable
customers to access and compare offers not only within Tunisia but also across different
countries. By adding localization, users will receive more relevant and region-specific
information, improving the overall user experience.
Moreover, integrating operators from around the globe will provide a comprehensive
database, allowing for more extensive market analysis and insights. This global perspective
will benefit telecom operators by offering a broader understanding of international market
trends and customer preferences, thus enabling them to adapt their strategies and services
accordingly.