HunFlair - Data Sets

Here you can find an overview about biomedical NER data sets integrated in HunFlair.

Content: Overview | HUNER Data Sets | BioBERT Evaluation Splits

Overview

HunFlair integrates 31 biomedical named entity recognition (NER) data sets and provides them in an unified format to foster the development and evaluation of new NER models. All data set implementations can be found in flair.datasets.biomedical.

Corpus	Data Set Class	Entity Types	Reference
AnatEM	ANAT_EM	Anatomical entities	Paper, Website
Arizona Disease	AZDZ	Disease	Website
BioCreative II GM	BC2GM	Gene	Paper
BioCreative V CDR task	CDR	Chemical, Disease	Paper, Website
BioInfer	BIO_INFER	Gene/Protein	Paper
BioNLP'2013 Cancer Genetics (ST)	BIONLP2013_CG	Chemical, Disease, Gene/Protein, Species	Paper
BioNLP'2013 Pathway Curation (ST)	BIONLP2013_PC	Chemical, Gene/Proteins	Paper
BioSemantics	BIOSEMANTICS	Chemical, Disease	Paper, Website
CellFinder	CELL_FINDER	Cell line, Gene, Species	Paper
CEMP	CEMP	Chemical	Website
CHEBI	CHEBI	Chemical, Gene, Species	Paper
CHEMDNER	CHEMDNER	Chemical	Paper
CLL	CLL	Cell line	Paper
DECA	DECA	Gene	Paper
FSU	FSU	Gene	Paper
GPRO	GPRO	Gene	Website
CRAFT (v2.0)	CRAFT	Chemical, Gene, Species	Paper
CRAFT (v4.0.1)	CRAFT_V4	Chemical, Gene, Species	Website
GELLUS	GELLUS	Cell line	Paper
IEPA	IEPA	Gene	Paper
JNLPBA	JNLPBA	Cell line, Gene	Paper
LINNEAUS	LINNEAUS	Species	Paper
LocText	LOCTEXT	Gene, Species	Paper
miRNA	MIRNA	Disease, Gene, Species	Paper
NCBI Disease	NCBI_DISEASE	Disease	Paper
Osiris v1.2	OSIRIS	Gene	Paper
Plant-Disease-Relations	PDR	Disease	Paper, Website
S800	S800	Species	Paper
SCAI Chemicals	SCAI_CHEMICALS	Chemical	Paper
SCAI Disease	SCAI_DISEASE	Disease	Paper
Variome	VARIOME	Gene, Disease, Species	Paper

_{Note: The table just gives an overview about the entity types of the individual corpora. Please refer to the original publications for annotation details.}

HUNER Data Sets

Next to the integration of the biomedical data sets, HunFlair provides the fixed splits used by HUNER (Weber et al.) to improve comparability of evaluations

Entity Type	Data Set Class	Contained Data Sets
Cell Line	HUNER_CELL_LINE	HUNER_CELL_LINE_CELL_FINDER, HUNER_CELL_LINE_CLL, HUNER_CELL_LINE_GELLUS, HUNER_CELL_LINE_JNLPBA
Chemical	HUNER_CHEMICAL	HUNER_CHEMICAL_CDR, HUNER_CHEMICAL_CEMP, HUNER_CHEMICAL_CHEBI, HUNER_CHEMICAL_CHEMDNER, HUNER_CHEMICAL_CRAFT_V4, HUNER_CHEMICAL_SCAI
Disease	HUNER_DISEASE	HUNER_DISEASE_CDR, HUNER_DISEASE_MIRNA, HUNER_DISEASE_NCBI, HUNER_DISEASE_SCAI, HUNER_DISEASE_VARIOME
Gene/Protein	HUNER_GENE	HUNER_GENE_BC2GM, HUNER_GENE_BIO_INFER, HUNER_GENE_CELL_FINDER, HUNER_GENE_CHEBI, HUNER_GENE_CRAFT_V4, HUNER_GENE_DECA, HUNER_GENE_FSU, HUNER_GENE_GPRO, HUNER_GENE_IEPA, HUNER_GENE_JNLPBA, HUNER_GENE_LOCTEXT, HUNER_GENE_MIRNA, HUNER_GENE_OSIRIS, HUNER_GENE_VARIOME
Species	HUNER_SPECIES	HUNER_SPECIES_CELL_FINDER, HUNER_SPECIES_CHEBI, HUNER_SPECIES_CRAFT_V4, HUNER_SPECIES_LINNEAUS, HUNER_SPECIES_LOCTEXT, HUNER_SPECIES_MIRNA, HUNER_SPECIES_S800, HUNER_SPECIES_VARIOME

BioBERT evaluation splits

To ease comparison with BioBERT, HunFlair provides the splits used by Lee et al.: BIOBERT_GENE_BC4CHEMD, BIOBERT_GENE_BC2GM, BIOBERT_GENE_JNLPBA, BIOBERT_CHEMICAL_BC5CDR, BIOBERT_DISEASE_BC5CDR, BIOBERT_DISEASE_NCBI, BIOBERT_SPECIES_LINNAEUS, and BIOBERT_SPECIES_S800

Note: To download and use the BioBERT corpora you need to install the package googledrivedownloader, since the files are hosted in Google Drive:

pip install googledrivedownloader