Twitter Analysis to Predict the Satisfaction of Saudi
Telecommunication Companies’ Customers

Abstract

The flexibility in mobile communications allows customers to quickly switch from one service provider to
another, making customer churn one of the most critical challenges for the data and voice telecommunication
service industry. In 2019, the percentage of post-paid telecommunication customers in Saudi Arabia
decreased; this represents a great deal of customer dissatisfaction and subsequent corporate fiscal losses.
Many studies correlate customer satisfaction with customer churn. The Telecom companies have depended
on historical customer data to measure customer churn. However, historical data does not reveal current
customer satisfaction or future likeliness to switch between telecom companies. Current methods of analysing
churn rates are inadequate and faced some issues, particularly in the Saudi market.
This research was conducted to realize the relationship between customer satisfaction and customer churn
and how to use social media mining to measure customer satisfaction and predict customer churn.
This research conducted a systematic review to address the churn prediction models problems and their
relation to Arabic Sentiment Analysis. The findings show that the current churn models lack integrating
structural data frameworks with real-time analytics to target customers in real-time. In addition, the findings
show that the specific issues in the existing churn prediction models in Saudi Arabia relate to the Arabic
language itself, its complexity, and lack of resources.
As a result, I have constructed the first gold standard corpus of Saudi tweets related to telecom companies,
comprising 20,000 manually annotated tweets. It has been generated as a dialect sentiment lexicon extracted
from a larger Twitter dataset collected by me to capture text characteristics in social media. I developed a
new ASA prediction model for telecommunication that fills the detected gaps in the ASA literature and fits
the telecommunication field. The proposed model proved its effectiveness for Arabic sentiment analysis and
churn prediction. This is the first work using Twitter mining to predict potential customer loss (churn) in
Saudi telecom companies, which has not been attempted before. Different fields, such as education, have
different features, making applying the proposed model is interesting because it based on text-mining.