A Retrieval-Augmented Generation Framework for Semantic Evaluation of Railway Education Curricula

The European rail sector is facing two major challenges: the loss of expertise associated with an ageing workforce and the changing competence demands created by digitalisation, automation, and data-driven systems. These challenges increase the need for railway higher education programmes to remain aligned with evolving industry requirements. However, curriculum evaluation in this area is still largely manual, time consuming, resource intensive making systematic comparison across programmes difficult, especially where similar competencies are expressed through different terminology. This paper presents the development and preliminary testing of an AI-assisted curriculum evaluation tool for railway higher education. The tool transforms unstructured curriculum documents into comparable representations and combines rule-based comparison with semantically informed analysis. The study adopts a design-based research approach and reports a comparative case study of two European Master's-level rail-related programmes, focusing on learning outcomes and subject areas. The preliminary test showed that the tool could extract and structure curriculum information from both programmes and support comparison at both syntactic and semantic levels. It identified patterns of alignment, partial overlap, and apparent gaps, including cases where conceptually related content was expressed differently across programmes. The study shows the feasibility of using such a tool to support more systematic curriculum evaluation in a specialised engineering domain. While the findings are preliminary, they indicate the potential of semantically informed curriculum evaluation as a decision-support approach for railway higher education.