Reproduce “Grading the Unannotated: Insights from the WorkRB Skill Extraction Challenge at RecSys-HR 2026” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Warre Veys, Luis Gasco, Jens-Joris Decorte
WorkRB replaces binary skill-extraction scoring with five benchmarks that grade candidate skills by relevance and evaluate the top 100 using nDCG@100. It averages results across four vacancy-sentence datasets and one skill-normalization dataset. Its winner scores 0.8458, versus 0.6292 for CurriculumMatch. These results show a large gap on this benchmark, not general skill understanding or deployment performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.