Evaluating restaurant service experience in Hue City via an ensemble sentiment analysis approach

Abstract

This study evaluates international tourists' experiences with restaurant services in Hue City using an ensemble sentiment analysis approach. By combining VADER, Cardiff EN, and XLM-RoBERTa models to process a dataset of 39,465 English online reviews, the research records an inter-model agreement rate of 90.53%. Overall, customer experience reflects an outstanding level of satisfaction, with 37,816 positive reviews, driven primarily by the quality of the food and the service attitude. Further analysis reveals that the "Family" group exhibits the highest level of satisfaction, whereas the "Couples" group, despite constituting the largest customer segment, records a lower average rating. Based on these empirical findings, the study proposes practical managerial implications for culinary businesses: proactively establishing monitoring processes for early warning signals from negative feedback and adjusting spatial design and service products to optimize experiences for each key target segment.