Abstract

The guest experience in the hotel sector is increasingly influenced by online opinions, which represent a valuable source for analyzing customer satisfaction and expectations. This study presents an applied test on the cases of Rome and Milan, two of Italy’s main tourist hubs, through the systematic analysis of reviews published on Booking.com. Sentiment analysis methods based on DistilBERT and topic modeling using Latent Dirichlet Allocation (LDA) were employed to objectively identify the main sources of customer satisfaction and dissatisfaction. The dataset, consisting of 52.033 reviews, underwent a rigorous pre-processing phase, including translation, normalization, and lemmatization of the texts. Sentiment analysis revealed that 60,4% of cases were positive, 37,7% negative, and 1,9% neutral. The words most commonly associated with positive judgments were “room”, “breakfast”, “staff”, “location”, “clean”, “comfortable”, “friendly”, and “helpful”—terms reflecting appreciation for the quality of rooms, breakfast, the professionalism of the staff, and the location of the hotels. Topic modeling on negative reviews, on the other hand, highlighted recurring critical issues related to “room”, “small”, “bathroom”, “shower”, “noise”, “price”, and “staff”. These words indicate dissatisfaction regarding the size and comfort of rooms, maintenance problems, breakfast quality, noise, and value for money. The results provide empirical evidence of both the strengths and critical points perceived by guests, offering practical tools for improving the hotel offering in highly competitive urban destinations.

Download

Share

COinS