Relation extraction dataset compiled from PubMed Central fulltext articles referenced in the BRENDA database

documents.json - TinyDB document database containing PubMed references cited by BRENDA, along with their respective mentions of enzymes, bacteria, bacterial strains, or other organisms, and with relations between these entities, as annotated by the BRENDA curators. documents.schema.json - JSON schema describing documents.json training_data.csv, validation_data.csv, test_data.csv - Subset of articles from documents.json that are available as fulltext via PubMed Central and split into training, validation, and test. pmc_linguistics_articles.json - Articles from the field of psycholinguistics, extracted from PubMed Central, to serve as helpful noise in model training.