Scholarlink

<p>This study examines 5,740 academic articles from Northeast Asia journals spanning 2000–2020, employing text mining, network centrality measures, topic modeling, and dynamic topic analysis. Results reveal significant changes in degree, closeness, and eigenvector centrality, and shifts in topic prevalence across the two decades. To assess topic similarity between LDA-generated and expert-labeled topics, the study uses Jaccard and Cosine similarity metrics, incorporating saliency, relevance, and term frequency. The similarity remained relatively stable over time. Statistical significance of topic trends was tested using the Mann-Kendall Trend Test and Chi-Square Test, revealing a significant increase in economic topics over political ones. This integrated approach offers insights into evolving interests and research dynamics in Northeast Asia, beyond the scope of traditional economic and political analyses.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning LDA-Generated and Expert-Labeled Topic Analysis Using Saliency and Relevance for Northeast Asia

  • 期刊论文

  • 发表日期：

    2026年3月24日

查看全文
  • Chae-Deug Yi

摘要

This study examines 5,740 academic articles from Northeast Asia journals spanning 2000–2020, employing text mining, network centrality measures, topic modeling, and dynamic topic analysis. Results reveal significant changes in degree, closeness, and eigenvector centrality, and shifts in topic prevalence across the two decades. To assess topic similarity between LDA-generated and expert-labeled topics, the study uses Jaccard and Cosine similarity metrics, incorporating saliency, relevance, and term frequency. The similarity remained relatively stable over time. Statistical significance of topic trends was tested using the Mann-Kendall Trend Test and Chi-Square Test, revealing a significant increase in economic topics over political ones. This integrated approach offers insights into evolving interests and research dynamics in Northeast Asia, beyond the scope of traditional economic and political analyses.