Public Transit Customer Satisfaction Dimensions Discovery from Online Reviews

Abstract

Online user-generated content provides a valuable source for identifying dimensions of services. This study proposes a framework for extracting the dimensions of consumer satisfaction of public transportation services using unsupervised latent Dirichlet allocation model. A pilot study was performed on 17,747 online user reviews collected from 1452 public transportation agencies (including streetcar, light rail, heavy rail, boat, and aerial tram) in the United States over 8 years. The proposed approach is able to identify a few dimensions that were not discussed in the previous literature. This research also provides an alternative method to collectively gather users’ feedback and efficiently pre-process textual data related to transit customer satisfaction.