The Velocity of Relevance: Mapping the Structural Divergence Between Labor Market Signals and University Curricula in Indonesia via Text Mining and Network Analysis

Abstract

The persistent disconnect between higher education outcomes and labor market demands, frequently termed the skills mismatch, remains a critical barrier to Indonesia's economic competitiveness in the Fourth Industrial Revolution. Traditional survey-based methodologies often lack the granularity to capture dynamic market shifts and technical nuances.  This study employs a Big Data approach, utilizing automated web scraping to harvest N = 1,042,500 unique job advertisements from major Indonesian portals and N = 4,500 course syllabi from 50 top-tier Indonesian universities between 2023 and 2024. We applied Natural Language Processing, specifically Latent Dirichlet Allocation for topic modeling, and Social Network Analysis to calculate semantic overlap and centrality measures between industry demands and academic provision. We utilized the Overlap Coefficient to correct for corpus size imbalance. The analysis reveals a structural divergence: while 82% of job ads prioritize Digital Fluency and Agile Project Management, only 28% of curricula explicitly integrate these competencies. Network analysis identifies Data Analysis as a peripheral node in academic graphs but a central hub in industry networks with a Betweenness Centrality of 0.45. Conversely, theoretical constructs dominant in academia show weak linkage to employability clusters. In conclusion, the findings evidence a systemic velocity gap where industry requirements evolve three times faster than curriculum adaptation. We propose a dynamic, API-driven curriculum model to mitigate this asymmetry.