MedDeID Dutch synthetic corpus, synthetic benchmark and annotation guidelines
收藏官方服务：
资源简介：
The MedDeID synthetic Dutch model-development corpus (6,493 documents), fixed physician-reviewed synthetic benchmark (300 documents), and bilingual ProductionLabels_v1 annotation guidelines. Dataset records use canonical JSONL with primary annotations only in spans, benchmark subannotations nested under their owning span, and the canonical patient and caregivers metadata keys. The resources contain no real patient notes or personal information.
提供机构：
Zenodo创建时间：
2026-08-18
