Using Deep Learning Networks to Predict Telecom Company Customer Satisfaction Based on Arabic Tweets
Abstract
Information systems are transforming businesses, which are using modern technologies towards new business models based on digital solutions, which ultimately lead to the design of novel socio-economic systems. Sentiment analysis is, in this context, a thriving research area. This paper is a case study of Saudi telecommunications (telecom) companies, using sentiment analysis for customer satisfaction based on a corpus of Arabic tweets. This paper compares, for the first time for Saudi social media in telecommunication, the most popular machine learning approach, support vector machine (SVM), with two deep learning approaches: long short-term memory (LSTM) and gated recurrent unit (GRU). This study used LSTM and GRU with two different implementations, adding attention mechanism and character encoding. The study concluded that the bidirectional-GRU with attention mechanism achieved a better performance in the telecommunication domain and allowed detection of customer satisfaction in the telecommunication domain with high accuracy.
Paper Type
Event
Using Deep Learning Networks to Predict Telecom Company Customer Satisfaction Based on Arabic Tweets
Information systems are transforming businesses, which are using modern technologies towards new business models based on digital solutions, which ultimately lead to the design of novel socio-economic systems. Sentiment analysis is, in this context, a thriving research area. This paper is a case study of Saudi telecommunications (telecom) companies, using sentiment analysis for customer satisfaction based on a corpus of Arabic tweets. This paper compares, for the first time for Saudi social media in telecommunication, the most popular machine learning approach, support vector machine (SVM), with two deep learning approaches: long short-term memory (LSTM) and gated recurrent unit (GRU). This study used LSTM and GRU with two different implementations, adding attention mechanism and character encoding. The study concluded that the bidirectional-GRU with attention mechanism achieved a better performance in the telecommunication domain and allowed detection of customer satisfaction in the telecommunication domain with high accuracy.
Recommended Citation
Almuqren, L. A. R., Qasem, M. M. & Cristea, A. I. (2019). Using Deep Learning Networks to Predict Telecom Company Customer Satisfaction Based on Arabic Tweets. In A. Siarheyeva, C. Barry, M. Lang, H. Linger, & C. Schneider (Eds.), Information Systems Development: Information Systems Beyond 2020 (ISD2019 Proceedings). Toulon, France: ISEN Yncréa Méditerranée.