Emerging Skills Temporal Knowledge Graphs from Job Postings (2000–2024)

This dataset accompanies the ECAI 2025 submission titled:"Early Detection of Skills Demand with Deep Learning-based Timeline Cover on Temporal Knowledge Graphs" It contains:- A structured collection of 250 annotated job postings in the Data Science domain (2000–2024) across five key sectors (Banking, Insurance, Consultancy, Pharma, Retail)- Manually constructed ground truth intervals for skill relevance, distinguishing emerging from established competencies The dataset was used to evaluate a novel, unsupervised early skills detection framework, based on the MinTCover problem on temporal graphs. Evaluation metrics include skill extraction precision/recall, Intersection-over-Union for activation intervals, and detection delay vs baseline methods. All data has been anonymized and aggregated to comply with privacy requirements. Licensed under Creative Commons Attribution 4.0 International. DOI for associated paper will be added upon acceptance. Please his dataset is intended for research purposes only. Any references to real companies, job advertisements are purely coincidental and do not imply any real-world affiliation or endorsement.