Mining Customer Opinion for Topic Modeling Purpose: Case Study of Ride-Hailing Service Provider

Abstract

The popularity of ride-hailing services in the form of smartphone application as a transportation solution has become center of attention. The convenience offered has made many people use it in daily life and discuss it on social media. As a result, ride-hailing service providers utilize social media for capturing customers' opinions and marketing their services. If customers' statements about ride-hailing services are analyzed further, service providers can get insight for evaluating their services to meet customers' satisfaction. Text mining approach can be useful to analyze large number of posts and various writing styles to extract hidden information. Furthermore, by applying topic modeling, service providers can identify the important points that were spoken by customers without previously giving label or category to the text. Latent Dirichlet Allocation was used in this study to extract topics based on the posts from ride-hailing customers published on Twitter. This study used 40 parameter combinations for LDA to get the best one to obtain the topics. Based on the perplexity value, there were 9 topics discussed by customers in their posts including the top words in each topic. The output of this study can be used for the service providers to evaluate and improve the services.