MedSETNER: A benchmark corpus for extracting dataset names from medical scientific literature

MedSETNER is the first annotated benchmark for extracting dataset names from medical and biomedical research articles. Dataset references in medical papers tend to be embedded in free-form prose, written with inconsistent nomenclature, and rarely tied to a persistent identifier. Existing dataset-mention corpora cover computer science or general science publications and do not cover medical text; that is the gap this corpus fills.