Enhancing hotel consumer recommendation decisions: A rough set approach for predictive analysis of online reviews

In today's digital age, online reviews have become a crucial resource for travelers who are trying to decide on a hotel to book. For hotel owners, being able to predict hotel recommendations based on these reviews holds great importance. It helps them understand what customers prefer, enhance their services, and attract more potential guests. In this research article, we introduce a novel approach that uses a concept called rough set theory to make accurate predictions about hotel recommendations using online review data. This method involves selecting important features and making decisions based on rough set theory. By doing so, we aim to uncover the key aspects that influence hotel recommendations. This can provide hotel managers with valuable insights to better meet guest expectations and boost their satisfaction. Our study uses a framework built upon rough set theory to perform sentiment analysis. We focus on a dataset of text reviews from hotels, where each review is transformed into a pair of characteristics during the data preparation phase. We then use rough set theory to explore the rules, main attributes, and essential features that can help categorize these hotel text reviews for future recommendations. We double-checked for duplicate or unnecessary attributes and compared our results with popular algorithms to ensure robustness. Our experiments clearly indicate that using rough set theory for making predictive recommendations is the most effective choice. It achieves impressive accuracy (97.00%) and allows for swift classification even when dealing with massive amounts of data. • A rough set-based approach is proposed for hotel recommendation. • The method improves accuracy by extracting decision rules from reviews. • Achieves 97.00% accuracy on real-world hotel review data. • Offers interpretable and transparent recommendations. • Demonstrates strong performance on large-scale sentiment data.