Research Hotspots and Evolutionary Trends ofEarthquake-Induced Landslides: A BibliometricAnalysis Based on Web of Science

Abstract

Earthquake-induced landslides are a major natural hazard and have attracted extensive global attention. Using 9,160 articles retrieved from the Web of Science Core Collection, we screened 6,956 publications highly relevant to earthquake-induced landslides through Python-based processing and manual verification. We combined VOSviewer with a latent Dirichlet allocation (LDA) topic model to quantify publication output, map keyword networks, delineate thematic structure, and trace research hotspot evolution. Publication trends exhibit three stages: an initial stage (1972–1999), a steady-growth stage (2000–2008), and a rapid-growth stage (2009–2024). The global research landscape has shifted from early dominance by Europe and North America to a multipolar pattern, with China emerging as a leading contributor while the United States, Italy, and Japan remain consistently productive. Research hotspots evolved from fundamental geological issues and slope-stability mechanisms to the widespread use of GIS and remote sensing for landslide inventory construction, spatial mapping, and hazard monitoring, followed by rapid growth in numerical modeling and machine learning. The LDA results cluster the literature into six research directions, and topic similarity analysis suggests that “mechanism studies” are most tightly coupled with “dynamic modeling,” whereas “cascading hazard impact assessment” is relatively independent. Overall, future research in this field is likely to emphasize (1) developing high-quality landslide inventories; (2) improving model transferability across events and regions; and (3) promoting application-oriented translation and decision support for risk governance.