Research Trends and Challenges in Diagnostic Classification Models: Insights from Dynamic Topic Modeling

This study aims to provide a comprehensive understanding of the research landscape of Diagnostic Classification Models (DCMs) and to support future model development and application by systematically reviewing research trends from 1992 to 2024. DCMs have received growing attention for their characteristic of delivering fine-grained diagnostic information on attribute mastery. However, despite decades of research, systematic analyses of trends in DCMs have remained limited, making it difficult to define the research scope and anticipate future developments. To address this gap, Dynamic Topic Modeling (DTM) was employed to explore key research topics and their temporal evolution. Among the identified topics, 13 were labeled, highlighting prominent themes such as Cognitive Diagnostic Computerized Adaptive Testing (CD-CAT), attribute hierarchy modeling, and Bayesian methodologies. Recently, the rise of artificial intelligence, especially deep learning, has accelerated research on neural network-based DCMs. By providing a comprehensive overview of DCMs research trends, this study assessed progress in the field and provided a foundation for future model development and application.