Skip to content

Tools for curating biomedical training data for large-scale language modeling

Notifications You must be signed in to change notification settings

ylaboratory/biomedical

 
 

Repository files navigation

BigBIO: Biomedical Dataset Library

BigBIO (BigScience Biomedical) is an open library of biomedical dataloaders built using Huggingface's (🤗) datasets library for data-centric machine learning.

Our goals include:

  • Lightweight, programmatic access to biomedical datasets at scale
  • Promoting reproducibility in data processing
  • Better documentation for dataset provenance, licensing, and other key attributes
  • Easier generation of meta-datasets for natural language prompting, multi-task learning

Currently BigBIO provides support for:

  • 126+ biomedical datasets
  • 10+ languages
  • 12 task categories
  • Harmonized dataset schemas by task type
  • Metadata on licensing, coarse/fine-grained task types, domain, and more!

How to Use BigBIO

The preferred way to use these datasets is to access them from the Official BigBIO Hub.

Minimally, ensure you have the datasets library installed. Preferably, install the requirements as follows:

pip install -r requirements.txt.


You can access BigBIO datasets as follows:

from datasets import load_dataset
data = load_dataset("bigbio/biosses")

In most cases, scripts load the original schema of the dataset by default. You can also access the BigBIO split that streamlines access to key information in datasets given a particular task.


For example, the biosses dataset follows a pairs based schema, where text-based inputs (sentences, paragraphs) are assigned a "translated" pair.

from datasets import load_dataset
data = load_dataset("bigbio/biosses", name="biosses_bigbio_pairs")

Generally, you can load your datasets as follows:

# Load original schema
data = load_dataset("bigbio/<your_dataset>")

# Load BigBIO schema
data = load_dataset("bigbio/<your_dataset_here>", name="<your_dataset>_bigbio_<schema_name>")

Check the datacards on the Hub to see what splits are available to you. You can find more information about schemas in Documentation below.

Benchmark Support

BigBIO includes support for almost all datasets included in other popular English biomedical benchmarks.

Task Type Dataset BigBIO (ours) BLUE BLURB BoX DUA needed
NER BC2GM
NER BC5-chem
NER BC5-disease
NER EBM PICO
NER JNLPBA
NER NCBI-disease
RE ChemProt
RE DDI
RE GAD
QA PubMedQA
QA BioASQ
DC HoC
STS BIOSSES
STS MedSTS *
NER n2c2 2010
NER ShARe/CLEF 2013 *
NLI MedNLI
NER n2c2 deid 2006
DC n2c2 RFHD 2014
NER AnatEM
NER BC4CHEMD
NER BioNLP09
NER BioNLP11EPI
NER BioNLP11ID
NER BioNLP13CG
NER BioNLP13GE
NER BioNLP13PC
NER CRAFT *
NER Ex-PTM
NER Linnaeus
POS GENIA *
SA Medical Drugs
SR COVID private
SR Cooking private
SR HRT private
SR Accelerometer private
SR Acromegaly private

* denotes dataset implementation in-progress

Documentation

Tutorials

TBA - Links may not be applicable yet!

Contributing

BigBIO is an open source project - your involvement is warmly welcome! If you're excited to join us, we recommend the following steps:

  • Looking for ideas? See our Volunteer Project Board to see what we may need help with.

  • Have your own idea? Contact an admin in the form of an issue.

  • Implement your idea following guidelines set by the official contributing guide

  • Wait for admin approval; approval is iterative, but if accepted will belong to the main repository.

Currently, only admins will be merging all accepted changes to the Hub.

Feel free to join our Discord!

Citing

If you use BigBIO in your work, please cite

@article{fries2022bigbio,
	title = {
		BigBIO: A Framework for Data-Centric Biomedical Natural Language
		Processing
	},
	author = {
		Fries, Jason Alan and Weber, Leon and Seelam, Natasha and Altay,
		Gabriel and Datta, Debajyoti and Garda, Samuele and Kang, Myungsun
		and Su, Ruisi and Kusa, Wojciech and Cahyawijaya, Samuel and others
	},
	journal = {arXiv preprint arXiv:2206.15076},
	year = 2022
}

Acknowledgements

BigBIO is a open source, community effort made possible through the efforts of many volunteers as part of BigScience and the Biomedical Hackathon.

About

Tools for curating biomedical training data for large-scale language modeling

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 93.8%
  • HTML 5.2%
  • Other 1.0%