User Reviews Sentiment Analysis on Internet Service Provider in Indonesia using Bidirectional Recurrent Neural Network

The COVID-19 pandemic in 2020 led to a significant increase in internet use, a trend that persisted post-pandemic due to hybrid work models and digital advancements. In Indonesia, where the digital economy is projected to reach USD 110 billion by 2025, reliable internet service is vital for the burgeoning digital economy. Sentiment analysis on user reviews provides critical insights into customer satisfaction and areas for improvement, directly benefiting ISPs by guiding service enhancements. This study employed sentiment analysis on the MyIndiHome app which was developed by Indonesia's leading ISP, IndiHome. MyIndiHome reviews from the Google Play Store can be utilized to assess user satisfaction. By utilizing Bidirectional Long Short-Term Memory (Bi-LSTM) and Bidirectional Gated Recurrent Unit (Bi-GRU) models, the research classified the sentiment of these reviews into positive or negative categories. The models' performance was compared to identify the most effective approach for this task. The study aimed to recommend the best model for sentiment analysis, aiding in enhancing ISP services and supporting Indonesia's digital growth. We found that Bi-GRU was the best of both models, offering better performance with 95.87% accuracy on testing data and also came with faster training time. With this high accuracy, Bi-GRU could be reliably used to conduct sentiment analysis on MylndiHome reviews.