17:15 〜 18:45
[MIS23-P12] Classifying and Trend Analyzing Geological Research Themes in South Korea Using Topic Models and Statistical Methods
キーワード:Topic mdoeling, Trend analysis, Latent Dirichlet Allocation, Statistical methods
The field of geology in South Korea has significantly evolved and expanded through interdisciplinary research since the mid-20th century. This research aims to identify the key topics in South Korean geological studies, analyze their trends over time, and categorize them based on their topic relationships. By applying latent Dirichlet allocation (LDA) to 9,966 articles from 10 Korean geological journals published between 1964 and 2021, we identified 18 distinct topics. Our methodology involved 1) tracking the trend of these topics to classify them as trending upwards (n=10), downwards (n=3), or remaining consistent (n=5), 2) assessing the breadth of each topic as either broad (n=14) or narrow (n=4), 3) organizing the topics into four clusters based on Euclidean distance, and 4) mapping a network to depict their relationships. The results of trend analysis showed that ‘Economic geology’, ‘Petrology’, and ‘Stratigraphy’ were the main focus areas until 1996. After this period, ‘Environmental geology’ and ‘Hydrogeology’ rose in importance, emerging as critical areas. Among the identified clusters (engineering, environment, field survey, and chemistry), the engineering cluster showed the most robust connections between topics (mean weight = 5.18), indicating an extensive use of engineering methods in addressing geological issues in South Korea. This study offers insights into the dynamic shifts in geological research topics, providing a foundation for further research and collaboration among geology scholars.
Acknowledgements: This research was supported by Learning & Academic research institution for Master’s·PhD students, and Postdocs(LAMP) Program of the National Research Foundation of Korea(NRF) grant funded by the Ministry of Education(No.RS-2023-00301702)
Acknowledgements: This research was supported by Learning & Academic research institution for Master’s·PhD students, and Postdocs(LAMP) Program of the National Research Foundation of Korea(NRF) grant funded by the Ministry of Education(No.RS-2023-00301702)