Beyond Star Ratings and Scores: Unveiling Guests' "Voices" in AI Hotel Online Reviews via LDA

With the in-depth integration of artificial intelligence (AI) technology and the hotel industry, AI-smart hotels have developed rapidly. However, issues such as weakened emotional connections caused by technological substitution and polarized user reviews have become increasingly prominent. Nevertheless, there is a lack of in-depth analytical research on online user reviews of smart hotels, making it an urgent challenge to accurately grasp user needs. This study employs the Latent Dirichlet Allocation (LDA) topic model to conduct text mining on online reviews of AI-smart hotels. Through steps including data preprocessing, model construction, and optimization, it extracts 8 core topics from the reviews, covering dimensions such as smart facilities, overall evaluation, repurchase intention, and environmental hygiene, which comprehensively reflect users' key concerns. Innovatively, this research distills consumer-focused topics from massive online reviews of AI hotels, addressing the problem of identifying consumers' needs for AI-smart hotels amid complex textual information and filling the gap in relevant theoretical research. Additionally, it provides operational and referential suggestions from three perspectives: AI hotel marketers, AI technology manufacturers, and consumers.