ESCOX: A tool for skill and occupation extraction using LLMs from unstructured text

ESCOX, also known as ESCOSkillExtractor, is an open-source, non-proprietary tool for identifying and classifying skills, skillsets, and occupations from job postings and general text. It utilizes the European Skills, Competences, Qualifications and Occupations (ESCO) taxonomy to structure extraction...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Software impacts Jg. 25; S. 100772
Hauptverfasser: Kavargyris, Dimitrios Christos, Georgiou, Konstantinos, Papaioannou, Eleanna, Petrakis, Konstantinos, Mittas, Nikolaos, Angelis, Lefteris
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier B.V 01.07.2025
Schlagworte:
ESCO
Job vacancies
Large language model
Skill & occupation extraction
Large language model
ESCO
Job vacancies
Skill & occupation extraction
ISSN:2665-9638, 2665-9638
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:ESCOX, also known as ESCOSkillExtractor, is an open-source, non-proprietary tool for identifying and classifying skills, skillsets, and occupations from job postings and general text. It utilizes the European Skills, Competences, Qualifications and Occupations (ESCO) taxonomy to structure extraction, addressing the need for taxonomy-aligned skill identification in unstructured labor market data. Developed within the SKILLAB EU Horizon project, ESCOX combines LLMs and text embeddings to map content to standardized categories. It offers a user-friendly graphical interface for researchers, educators, and HR professionals, supporting skills gap analysis, training, recruitment, and policy planning, and contributing to the development of a skills-based economy. [Display omitted] •Identifying relevant skills is essential for shaping the modern workforce.•ESCOX is an open-source tool for skill and occupation extraction using the ESCO taxonomy.•High-speed extraction enabled by precomputed embeddings and LLMs.•User-friendly interface supports use without requiring coding knowledge.•Supports skill extraction across domains with flexible preprocessing options.
ISSN:2665-9638
2665-9638
DOI:10.1016/j.simpa.2025.100772