language:
- en
size_categories:
- 100K<n<1M
task_categories:
- text-classification
pretty_name: BioRel
dataset_info:
features:
- name: text
dtype: string
- name: relation
dtype: string
- name: h
struct:
- name: id
dtype: string
- name: name
dtype: string
- name: pos
sequence: int64
- name: t
struct:
- name: id
dtype: string
- name: name
dtype: string
- name: pos
sequence: int64
splits:
- name: train
num_bytes: 179296923
num_examples: 534277
- name: validation
num_bytes: 38273878
num_examples: 114506
- name: test
num_bytes: 38539441
num_examples: 114565
download_size: 107508802
dataset_size: 256110242
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
tags:
- biology
- relation-classification
- medical
Dataset Card for BioRel
Dataset Description
- Repository: https://drive.google.com/drive/folders/1vw2zIxdSoqT2QALDbRVG6loLsgi2doBG
- Paper: BioRel: towards large-scale biomedical relation extraction
Dataset Summary
BioRel Dataset Summary:
BioRel is a comprehensive dataset designed for biomedical relation extraction, leveraging the vast amount of electronic biomedical literature available. Developed using the Unified Medical Language System (UMLS) as a knowledge base and Medline articles as a corpus, BioRel utilizes Metamap for entity identification and linking, and employs distant supervision for relation labeling. The training set comprises 534,406 sentences, the validation set includes 218,669 sentences, and the testing set contains 114,515 sentences. This dataset supports both deep learning and statistical machine learning methods, providing a robust resource for training and evaluating biomedical relation extraction models. The original dataset is available here: https://drive.google.com/drive/folders/1vw2zIxdSoqT2QALDbRVG6loLsgi2doBG
We converted the dataset to the OpenNRE format using the following script: https://github.com/GDAMining/gda-extraction/blob/main/convert2opennre/convert_biorel2opennre.py
Languages
The language in the dataset is English.
Dataset Structure
Dataset Instances
An example of 'train' looks as follows:
{
"text": "algal polysaccharide obtained from carrageenin protects 80 to 100 percent of chicken embryos against fatal infections with the lee strain of influenza virus .",
"relation": "NA",
"h": {
"id": "C0032594",
"name": "polysaccharide",
"pos": [6, 20]
},
"t": {
"id": "C0007289",
"name": "carrageenin",
"pos": [35, 46]
}
}
Data Fields
text
: the text of this example, astring
feature.h
: head entityid
: identifier of the head entity, astring
feature.pos
: character offsets of the head entity, a list ofint32
features.name
: head entity text, astring
feature.
t
: tail entityid
: identifier of the tail entity, astring
feature.pos
: character offsets of the tail entity, a list ofint32
features.name
: tail entity text, astring
feature.
relation
: a class label.
Citation
BibTeX:
@article{xing2020biorel,
title={BioRel: towards large-scale biomedical relation extraction},
author={Xing, Rui and Luo, Jie and Song, Tengwei},
journal={BMC bioinformatics},
volume={21},
pages={1--13},
year={2020},
publisher={Springer}
}
APA:
- Xing, R., Luo, J., & Song, T. (2020). BioRel: towards large-scale biomedical relation extraction. BMC bioinformatics, 21, 1-13.