CoDiet corpus

The CoDiet corpora created as part of a multi-institution collaboration to annotate biomedical literature. The repository contains 5 datasets (zipped) in BioC-JSON formats: CoDiet-Gold-public, 450 full-text documents independently annotated by 2 individuals and any disagreements adjudicated by a third independent person) CoDiet-Gold-private, 50 full-text documents unannotated to be used for benchmarking on Codabench (see link on GitHub) CoDiet-Electrum, 4,445 full-text documents machine-annotated (with post-processing rules) using the Gold-public labels/vocabulary only CoDiet-Silver, 4,445 full-text documents machine-annotated (with post-processing rules) using existing deep learning, rule-based, and dictionary-based methods CoDiet-Bronze, 4,445 full-text documents machine-annotated (without post-processing rules) using existing deep learning, rule-based, and dictionary-based methods If you use this resource, please cite the related works below.