Abstract:
It is important to highlight the significance of customer relationships for
telecommunication companies like Kcell. Hence, by providing customers with quick and
efficient channels to communicate and receive support, Kcell can improve customer
satisfaction and ultimately contribute to customer retention rate. This project serves as a
practical example of how state-of-the-art technologies, such as machine learning
algorithms and chatbots, can be implemented to optimize and improve customer service
in the telecommunication industry. In detail, the project is about the full development
process of a chatbot using a machine learning algorithm. The main purpose of the chatbot
is to assist clients of Kcell JSC with their questions related to the use of
telecommunication services and operations in a prompt manner. The frontend of the
chatbot is deployed on the Telegram platform and integrates to all platforms as Web,
Android, and iOS. The backend of the chatbot with an algorithm that uses machine
learning is deployed on Amazon Web Services EC2. The project includes the backend
with the design and implementation of a PostgreSQL database in Amazon Web Services
relational database service to store and manage the chatbot's backend data as well as user
feedback about the chatbot and chat logs. A machine learning model is trained on a
dataset obtained from a survey specifically designed for cellular operator users and from
artificially generated ones. Under-the-hood algorithm aims to provide an efficient and
user-friendly solution for telecommunication mobile questions, inquiries, and issues as in
Frequently Asked Questions (FAQ), ultimately improving the user experience and
increasing customer satisfaction due to the speed and accuracy of answers. Future work
of the project includes integration of chatbot functionality inside the Kcell mobile
application, expanding the dataset for machine learning algorithms to achieve better
training and testing in order to incorporate additional features and functionalities to
enhance the chatbot's capabilities.