A Machine Learning-Based Analysis of Tourism Recommendation Systems: Holistic Parameter Discovery and Insights

Abstract

Tourism is a cornerstone of the global economy, fostering cultural exchange and economic growth. As travelers increasingly seek personalized experiences, recommendation systems have become vital in guiding decision-making and enhancing satisfaction. These systems leverage advanced technologies such as IoT and machine learning to provide tailored suggestions for destinations, accommodations, and activities. This paper explores the transformative role of tourism recommendation systems (TRS) by analyzing data from 3,013 research articles published between 2000 and 2024 using a BERT-based methodology for semantic text representation and clustering. A robust software framework, integrating tools such as UMAP for dimensionality reduction and HDBSCAN for clustering, facilitated data modeling, cluster analysis, visualization, and the identification of key parameters in TRS. We discover a comprehensive taxonomy of 16 TRS parameters grouped into 4 macro-parameters. These include Personalized Tourism; Sustainability, Health and Resource Awareness; Adaptability & Crisis Management; and Social Impact & Cultural Heritage. These macro-parameters align with all three dimensions of the triple bottom line (TBL) -- social, economic, and environmental sustainability. The findings reveal key trends, highlight underexplored areas, and provide research-informed recommendations for developing more effective TRS. This paper synthesizes existing knowledge, identifies research gaps, and outlines directions for advancing TRS to support sustainable, personalized, and innovative travel solutions.