Trends in AI Research in the Field of Humanities: A Topic Modeling Approach Using LDA, BERTopic

Recent advancements in generative artificial intelligence have broadened the application of AI in various areas of the humanities, including language, the arts, and education. This study explores research trends in artificial intelligence within the humanities by applying topic modeling to 1,557 Korean abstracts collected from the RISS (Research Information Sharing Service) database. Two modeling techniques were employed: the traditional probabilistic method, Latent Dirichlet Allocation (LDA), and the neural embedding-based method, BERTopic. The results indicate that key themes such as ethics, education, religion, and culture were commonly identified by both models. BERTopic, in particular, revealed more fine-grained topic differentiation through its context-aware clustering approach. This study highlights the increasing integration of artificial intelligence into humanities research and contributes to understanding the evolving role of the humanities in the era of AI.