Bibliometric analysis of landslide research based on the WOS database

Abstract

This work, based on the Web of Science (WOS) database, collected 20,888 research articles published from 1982 to November 2021 on the topic of landslide(s). We performed a bibliometric analysis for the year, author, institution, country/region, foundation program, journal, and keywords. The abstracts in these articles are also analyzed in terms of the Latent Dirichlet Allocation (LDA) model in machine learning. Results show that the overall annual publication number is steadily increasing from 8 in 1982 to 2432 in 2020. Pradhan, Biswajeet, and Xu, Qiang are the authors with the most published papers, and Guzzetti, Fausto, and Dai, Fuchu are the authors with the highest average citations per article. The Chinese Academy of Sciences and the Italian Consiglio Nazionale Delle Ricerche are the institutions contributing the most articles. China and the United States contribute the highest article number, and Malaysia and Norway have the highest average citations per article. The journals Landslides and Engineering Geology host the most abundant landslide-related articles, while Geology and Environmental Geology have the most average citations per article. The number of articles funded by the Chinese National Natural Science Foundation far exceeds that of other funds. The keywords with high overall frequency are “model”, “GIS”, “susceptibility”, “earthquake”, “remote sensing”, “slope stability”, “deformation”, etc. The LDA model divides the articles into 19 topics, among which machine learning has the fastest growth rate. The top three topics in terms of frequency are ''risk assessment'', ''shallow landslides'' and ''evolution process and numerical simulation''. The keyword analysis is further clustered into four research areas, namely “Quaternary geology and geomorphology”, “landslide engineering geology”, “landslide database and risk assessment”, and “landslide monitoring and warning".