This paper introduces and applies a reader-centered analytical framework to examine how personality influences the interpretation of online hotel reviews. It addresses the gap between abundant sender-focused research and the limited attention to reader diversity, using an AI-based approach.
A five-step methodology was developed: (1) extraction of 997 English-language TripAdvisor reviews from nine five-star luxury hotels in Barcelona; (2) simulation of readers high in each Big Five personality trait through persona prompting; (3) automated scoring of each review by two large language models (GPT-4o and Claude 3 Haiku) on a continuous 1.00–5.00 scale; (4) creation of a results matrix combining human and simulated ratings; and (5) statistical and topic-based analysis, including bias, MAE, RMSE, correlation, LDA, and two-way ANOVA.
All simulated personas rated reviews slightly lower than human reviewers, with systematic differences between traits. Personality explained 22.8% of variance in ratings, compared with 2.6% for review topic. No significant trait–topic interaction was found.
The study focuses on luxury hotels in one destination and on English-language reviews. Future applications could explore other contexts, languages and AI models.
The methodology enables personality-aware monitoring of review interpretation, offering fine-grained insights for digital reputation management in hospitality.
Simulating diverse reader profiles supports more inclusive communication strategies and a better understanding of how digital content is perceived by different audiences.
This study proposes a structured, replicable methodology that combines personality theory, large language model simulation, and computational text analysis to capture heterogeneity in review reception within tourism research.
