Keywords :
Latent Dirichlet Allocation; Machine Learning; Travel satisfaction; TripAdvisor; urban metro systems; Classification models; Language processing; Latent Dirichlet allocation; Machine-learning; Metro system; Natural languages; Processing model; Tripadvisor; Urban metro system; Artificial Intelligence; Computer Science Applications; Information Systems and Management; Control and Optimization; Modeling and Simulation; Transportation
Abstract :
[en] As urban metro systems are the primary transportation mode used by tourists when exploring cities, it is imperative for public administrations to investigate the key factors that impact tourists' satisfaction towards these systems. This study uses structured and unstructured data from the TripAdvisor and TomTom Move platforms to predict tourism satisfaction perceptions toward 32 urban metro systems worldwide. The study utilized the Latent Dirichlet Allocation model to extract the number of dimensions from TripAdvisor reviews, trains a set of classification models to predict tourist satisfaction perceptions, and then applies the SHAP method to determine the features' impact on these satisfaction perceptions. Tourist satisfaction towards urban metro systems is influenced by the review title's mention of ease of use, accessibility, services, and facilities; the review content's mention of the system's utility; the city being in an Asian country; and the congestion levels. These findings provide a quick method for evaluating experiences and developing strategies for improvement.
Scopus citations®
without self-citations
1