Identifying Emerging Research Frontiers with Large Language Models: An Empirical Study for Engineering Management

Identifying emerging research frontiers is essential for tracking disciplinary developments and institutional strategic planning. However, existing methods for topic identification present several limitations, including insufficient semantic understanding, difficulty in reducing redundancy, and instability in generating and clustering topics without manual intervention. To address these challenges, we propose a systematic framework that integrates large language models (LLMs), a semantic embedding model, and quantitative indicator evaluation. Applying this framework to engineering management, we construct a delimited corpus of 350 synthesis-oriented articles from the Web of Science (WoS) and obtain standard topics ranked by a composite score incorporating frequency, centrality, and novelty scores. Then we carry out five duplicate experiments and successfully cluster eight major research directions from all the standard topics. The results are robustly tested, providing a solid evidence base for scientific management and data-driven policy making in this field. The proposed research framework not only supports engineering management research, but also offers a promising approach for identifying emerging research frontiers in other disciplines.