JobHop v2: A Large-Scale Career Trajectory Dataset from Unstructured Resumes

2026-07-13

AI Summary

Existing career trajectory datasets are limited in scale, often closed, or reliant on synthetic text, lacking large-scale public resources derived from real-world unstructured resumes. This work proposes an end-to-end large language model extraction framework that integrates inference-time control with a retry mechanism to efficiently extract structured career trajectories from approximately 440,000 anonymized multilingual resumes, achieving a 100% JSON parse rate. The resulting JobHop v2 dataset comprises 355,315 trajectories annotated with ESCO occupation codes, quarterly timestamps, and five-tier education levels. Comprehensive benchmark evaluations demonstrate its high quality, with the best-performing model approaching the inter-annotator agreement ceiling—falling short by only 1.1–2.7 percentage points.

Abstract

Large-scale, richly annotated career trajectory data underpins workforce planning, job recommendation, and labour market analysis, yet publicly available datasets are either small, closed to independent use, or built from pre-standardized occupational codes with LLM-synthesized rather than authentic free text. We present JobHop~v2, an improved version of the publicly available JobHop dataset, constructed through end-to-end large language model (LLM) extraction from a corpus of ${\sim}440{,}000$ pseudonymized, multilingual resumes provided by VDAB, the Flemish Public Employment Service. The released dataset comprises $355{,}315$ career trajectories annotated with ESCO occupational codes, quarter-level temporal information, and normalized five-level education attainment, broadening both the coverage and the annotation richness of the original release. Relative to v1, JobHop~v2 introduces a redesigned extraction pipeline based on reasoning-controlled LLM inference with a retry mechanism (achieving a 100% JSON parse rate), a richer extraction schema, and a revised evaluation protocol scored against three complementary annotation baselines. Evaluated against these baselines, our best extractor comes closest to the inter-annotator agreement ceiling among all compared models, trailing it by only 1.1-2.7 percentage points. The dataset and code are publicly released to support reproducible career-trajectory research.

Problem

• career trajectory
• unstructured resumes
• large-scale dataset
• workforce data
• occupational coding

Innovation

• reasoning-controlled LLM inference
• career trajectory extraction
• structured resume parsing
• ESCO occupational coding
• retry mechanism