Understanding User Perception of Ride-Hailing Services Sentiment Analysis and Topic Modelling using IndoBERT and BERTopic

Gojek is a unicorn startup that originated in Indonesia and disrupted the country’s transportation business ecosystem. They provide innovative solutions by enabling intercity transportation using a mobile app, thus filling the dire demand of most Indonesian for easy and relatively cheap rides. Listening to customers’ voices through social media is one of their strategies for developing product-market fit. Extracting meaningful insights or sentiments for the Indonesian language is challenging and previously done using a labeling-based method. However, the development of IndoBERT provides easier and better text mining methods using Indonesian language texts. We aim to understand Gojek’s customer perspective on service quality by performing sentiment analysis on their tweets containing their experiences when using Gojek’s mobile app. We employ IndoBERT in our model to classify the sentiments of 32.985 tweets, then extract the topics in each sentiment class to identify strengths and improvement opportunities in Gojek’s service dimension. The results are recommendations for Gojek to improve its service in several dimensions. Companies aiming to extract their customers’ voices can adopt the current study's model.