On developing diverse multiaspect hotel recommender systems based on TripAdvisor online reviews

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Purpose The primary objective of this study is to enhance hotel recommender systems by developing multiaspect methods that align closely with individual user preferences. The proposed multiaspect hybrid filtering (MAHF) and multiaspect content-based filtering (MACB) methods incorporate detailed review data and enable a deeper understanding of user preferences across multiple hotel attributes. Design/methodology/approach This study introduces two novel multiaspect recommender methods, MAHF and MACB, to improve hotel recommendation accuracy and enhance the understanding of user preferences. MAHF identifies users with similar preferences by comparing customer rating vectors (CRV), while MACB analyzes the similarity between CRVs and hotel rating vectors (HRV) to recommend hotels with matching attributes. Review data for hotels in five US cities were collected from TripAdvisor.com. Findings Experimental results showed that MAHF significantly outperformed benchmark models, demonstrating higher recommendation accuracy. The MACB method offers explainable recommendations, allowing users to understand why particular hotels are suggested. The study highlights the effectiveness of combining CRVs and HRVs and points to the potential for further enhancements by refining the definitions of CRVs and HRVs with additional factors. Originality/value This study advances hotel recommender systems by proposing MAHF and MACB, which apply detailed customer and hotel attribute vectors to the most widely adopted recommendation approaches: hybrid methods based on collaborative filtering (CF) and content-based methods. Both models use multiaspect filtering techniques to capture user preferences with greater precision while accurately identifying the distinctive characteristics of each hotel. They optimize recommendation accuracy and provide insights for hotel managers to establish more refined customer acquisition strategies. By demonstrating that multiaspect methods can facilitate more personalized hotel recommendations, this study makes a meaningful contribution to the ongoing advancement of hotel recommendation techniques.