Artificial Intelligence in Landslide Susceptibility: A Bibliometric Analysis
Berkant Konakoğlu, Sultan Sevinç KURT KONAKOĞLU
Abstract
ABSTRACT Landslides, a prevalent form of mass movement, represent significant natural disasters capable of inflicting substantial casualties and property damage. Determining landslide susceptibility is crucial for enhancing resilience against such events. Artificial intelligence (AI) methods have emerged as significant tools across geosciences. However, a comprehensive retrospective analysis of AI applications within landslide susceptibility modeling remains lacking. This study aims to address this gap by providing a detailed overview of AI utilization in landslide susceptibility research. Our research employed bibliometric methods to analyze a sample of 2892 studies published between 1991 and 2025, obtained from the Web of Science database. This bibliometric analysis identified China as the leading contributor to the field, both through national initiatives and contributions from its research institutions, such as China University of Geosciences. Prominent journals such as Geomorphology, Catena, Environmental Earth Sciences, Landslides, and Natural Hazards have played pivotal roles in advancing research within this domain. Notably, influential authors like Biswajeet Pradhan, Dieu Tien Bui, and Binh Thai Pham have made substantial contributions. The predominant methodological approach employed relies heavily on machine learning techniques and, more recently, deep learning architectures. Our keyword evolution analysis reveals a significant shift from machine learning methods toward eXplainable Artificial Intelligence approaches, which provide transparency in model decisions critical for hazard management applications. This comprehensive review will serve as a valuable resource for researchers investigating landslide susceptibility using AI and enable them to critically evaluate the evolution of scientific publications over time.