Over the past decade, Cedefop and Eurostat have collaborated to harness the potential of online job advertisements (OJAs) as a novel source of skills intelligence in Europe. This publication synthesises a decade of research and operational experience, documenting the development and evolution of a multilingual data production system (DPS) capable of extracting granular, near real-time insights into skills demand across European labour markets. The report details the methodologies used—from web scraping and natural language processing to machine learning and ontology-based classification—highlighting innovations in occupation and skills extraction, and the challenges of working with data not originally designed for statistical purposes.
Special focus is given to leveraging OJAs for monitoring the digital and green transitions, including the refinement of ESCO taxonomies through large language models and expert validation. The study also explores the extraction of information on qualifications and fields of study, enhancing understanding of labour market requirements beyond traditional occupational categories. Integration into Eurostat’s Web Intelligence Hub ensures methodological robustness and scalability, establishing OJA data as a vital complement to conventional statistics. This work not only demonstrates the feasibility and utility of big data in official statistics but also charts the way forward for responsive, evidence-based policy in a rapidly evolving labour market landscape.
Please log in or sign up to comment.