Структурний аналіз наукових кластерів у дослідженнях AIED: ієрархічна кластеризація тематичних моделей

Abstract

This study explores thematic structures in Artificial Intelligence in Education research through hierarchical clustering of topic models. Using data from Web of Science, Latent Dirichlet Allocation was applied to identify dominant research themes. Cosine similarity-based clustering revealed five key topics, including AI applications in medical education, machine learning in education, and large language models in learning. Future work will compare LDA with Non-negative Matrix Factorization and enhance visualization using MDS, t-SNE, and UMAP to refine topic differentiation and thematic interpretation.