Rapid changes in technologies and labor market demands have increased the concern regarding the compatibility between higher education and the skills required by employers. Existing studies often employ simple frequency counts and direct skill matching techniques to analyze the alignment between curriculum and job market. However, this approach provides limited insight into the structural relationships among the skills. Thus, this study aims to explore a different avenue of analysis by describing a methodology that integrates large language model (LLM)-based skill extraction with network analysis to examine the relationship between university curricula and job market requirements.

The methodology uses unstructured textual data from university course descriptions and online job advertisements. Skills are extracted and assigned relevance scores using an LLM, reflecting the extent to which a skill is taught within university programmes and demanded in the labor market, respectively. Based on these weighted skill representations, skill-document matrices are constructed and transformed into networks linking university programmes and job domains through shared skills in order to answer specific questions regarding the alignment of curriculum and job market.

Measures of overlap are quantified and visualized using heat maps, while network metrics; including degree centrality, betweenness centrality, closeness centrality, and eigenvector centrality, are employed to investigate the structural properties of the networks.

The proposed framework enables the identification of similarities and discrepancies between educational programmes and labor market demands, while also revealing deeper insight into the skills that occupy central positions within the networks. By integrating direct measures of skill overlap with network-based analyses, the methodology aims to provide an enhanced understanding of curriculum-labor market alignment than approaches based solely on frequency comparisons or isolated skill matching.

This study contributes a scalable methodology illustrated through its application on studying the relationship between higher education and labor market skill demands and showcases the potential of combining LLM-based skill extraction with network analysis to support curriculum evaluation and evidence-based educational policy making.