Mapping O*NET tasks to ISCO occupations using text similarity
Statistical Journal of the IAOS · 20.11.2026
DOI: 10.1177/18747655261507876 ↗Kurz gefasst
Fachartikel in der Zeitschrift Statistical Journal of the IAOS, online erschienen am 20.11.2026. Verfasst von Majid Einian. Die Arbeit nennt 14 Literaturangaben. Gefördert von: Stiftelsen Arcada.
Abstract (Originalsprache)
O*NET task data are widely used to measure technology exposure and skill demand, but O*NET is based on the U.S. Standard Occupational Classification (SOC) while international labour statistics use ISCO. Existing SOC–ISCO crosswalks operate at the occupation level, ignoring within-occupation task variation. This paper constructs a crosswalk from O*NET tasks to 4-digit ISCO-08 occupations using text similarity and ESCO information, covering all 63 O*NET releases from v4.0 to v30.3, with detailed validation across the 50 SOC 2010 and SOC 2018 releases (v15.1–v30.3). Across SOC 2018 releases, 18,796–19,281 tasks are assigned to 430–431 of 436 ISCO-08 groups, with 53.6–53.7% exact agreement and 83.0–83.1% major-group agreement against four existing SOC–ISCO-08 crosswalks. The crosswalk enables task-based measures of automation, skills, and AI exposure to be applied to internationally comparable labour statistics.
Förderung
- Stiftelsen Arcada
Zitieren
APA
Einian, M. (2026). Mapping O*NET tasks to ISCO occupations using text similarity. Statistical Journal of the IAOS. https://doi.org/10.1177/18747655261507876
Vancouver
Einian M. Mapping O*NET tasks to ISCO occupations using text similarity. Statistical Journal of the IAOS. 2026. doi:10.1177/18747655261507876
BibTeX
@article{Einian2026,
author = {Einian, Majid},
title = {Mapping O*NET tasks to ISCO occupations using text similarity},
journal = {Statistical Journal of the IAOS},
year = {2026},
doi = {10.1177/18747655261507876}
}
EUVIMED