Критерії аналітичної придатності тематичних моделей для аналізу текстових корпусів у дистанційній освіті

Abstract

The study examines the analytical suitability of topic modeling methods for exploring text corpora that reflect the use of artificial intelligence in distance education. It is shown that the effectiveness of topic modeling in this domain should be assessed not only by the ability to identify topics, but also by coherence, interpretability, robustness to parameter settings, and applicability to short and heterogeneous texts. The paper argues that the analytical value of topic models in this field depends on their ability to support both thematic structuring and the subsequent interpretation of results in educational and scientific contexts.