Hugging Face's logo

Datasets:
nlpaueb
/
biomrc

Languages:
English
Dataset card Files Community
4
biomrc / README.md
albertvillanova's picture
albertvillanova HF staff
Reorder split names (#1)
63c373d
preview code
|
raw
history blame
15.2 kB
metadata
language:
  - en
paperswithcode_id: biomrc
pretty_name: BIOMRC
dataset_info:
  - config_name: plain_text
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 1653301820
        num_examples: 700000
      - name: validation
        num_bytes: 119697683
        num_examples: 50000
      - name: test
        num_bytes: 147832373
        num_examples: 62707
    download_size: 408080356
    dataset_size: 1920831876
  - config_name: biomrc_large_A
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 1653301820
        num_examples: 700000
      - name: validation
        num_bytes: 119697683
        num_examples: 50000
      - name: test
        num_bytes: 147832373
        num_examples: 62707
    download_size: 408080356
    dataset_size: 1920831876
  - config_name: biomrc_large_B
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 1325877001
        num_examples: 700000
      - name: validation
        num_bytes: 96414040
        num_examples: 50000
      - name: test
        num_bytes: 118708586
        num_examples: 62707
    download_size: 343061539
    dataset_size: 1540999627
  - config_name: biomrc_small_A
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 206553549
        num_examples: 87500
      - name: validation
        num_bytes: 14957163
        num_examples: 6250
      - name: test
        num_bytes: 14807799
        num_examples: 6250
    download_size: 68879274
    dataset_size: 236318511
  - config_name: biomrc_small_B
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: train
        num_bytes: 165662937
        num_examples: 87500
      - name: validation
        num_bytes: 12047304
        num_examples: 6250
      - name: test
        num_bytes: 11911172
        num_examples: 6250
    download_size: 57706889
    dataset_size: 189621413
  - config_name: biomrc_tiny_A
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: test
        num_bytes: 70914
        num_examples: 30
    download_size: 22519
    dataset_size: 70914
  - config_name: biomrc_tiny_B
    features:
      - name: abstract
        dtype: string
      - name: title
        dtype: string
      - name: entities_list
        sequence: string
      - name: answer
        dtype: string
    splits:
      - name: test
        num_bytes: 59925
        num_examples: 30
    download_size: 19685
    dataset_size: 59925

Dataset Card for "biomrc"

Table of Contents

Dataset Description

Dataset Summary

We introduce BIOMRC, a large-scale cloze-style biomedical MRC dataset. Care was taken to reduce noise, compared to the previous BIOREAD dataset of Pappas et al. (2018). Experiments show that simple heuristics do not perform well on the new dataset and that two neural MRC models that had been tested on BIOREAD perform much better on BIOMRC, indicating that the new dataset is indeed less noisy or at least that its task is more feasible. Non-expert human performance is also higher on the new dataset compared to BIOREAD, and biomedical experts perform even better. We also introduce a new BERT-based MRC model, the best version of which substantially outperforms all other methods tested, reaching or surpassing the accuracy of biomedical experts in some experiments. We make the new dataset available in three different sizes, also releasing our code, and providing a leaderboard.

Supported Tasks and Leaderboards

More Information Needed

Languages

More Information Needed

Dataset Structure

Data Instances

biomrc_large_A

  • Size of downloaded dataset files: 389.18 MB
  • Size of the generated dataset: 1831.85 MB
  • Total amount of disk used: 2221.02 MB

An example of 'train' looks as follows.

This example was too long and was cropped:

{
    "abstract": "\"OBJECTIVES: @entity9 is a @entity10 that may result from greater occipital nerve entrapment. Entrapped peripheral nerves typica...",
    "answer": "@entity9 :: (MESH:D009437,Disease) :: ['unilateral occipital neuralgia']\n",
    "entities_list": ["@entity1 :: ('9606', 'Species') :: ['patients']", "@entity10 :: ('MESH:D006261', 'Disease') :: ['headache', 'Headache']", "@entity9 :: ('MESH:D009437', 'Disease') :: ['Occipital neuralgia', 'unilateral occipital neuralgia']"],
    "title": "Sonographic evaluation of the greater occipital nerve in XXXX .\n"
}

biomrc_large_B

  • Size of downloaded dataset files: 327.17 MB
  • Size of the generated dataset: 1469.61 MB
  • Total amount of disk used: 1796.78 MB

An example of 'train' looks as follows.

This example was too long and was cropped:

{
    "abstract": "\"BACKGROUND: Adults with physical disabilities are less likely than others to receive @entity2 screening. It is not known, howev...",
    "answer": "@entity2",
    "entities_list": ["@entity2", "@entity1", "@entity0", "@entity3"],
    "title": "Does a standard measure of self-reported physical disability correlate with clinician perception of impairment related to XXXX screening?\n"
}

biomrc_small_A

  • Size of downloaded dataset files: 65.69 MB
  • Size of the generated dataset: 225.37 MB
  • Total amount of disk used: 291.06 MB

An example of 'validation' looks as follows.

This example was too long and was cropped:

{
    "abstract": "\"PURPOSE: @entity120 ( @entity120 ) is a life-limiting @entity102 that presents as an elevated blood pressure in the pulmonary a...",
    "answer": "@entity148 :: (MESH:D001008,Disease) :: ['anxiety']\n",
    "entities_list": "[\"@entity1 :: ('9606', 'Species') :: ['patients']\", \"@entity308 :: ('MESH:D003866', 'Disease') :: ['depression']\", \"@entity146 :...",
    "title": "A predictive model of the effects of @entity308 , XXXX , stress, 6-minute-walk distance, and social support on health-related quality of life in an adult pulmonary hypertension population.\n"
}

biomrc_small_B

  • Size of downloaded dataset files: 55.03 MB
  • Size of the generated dataset: 180.84 MB
  • Total amount of disk used: 235.87 MB

An example of 'train' looks as follows.

This example was too long and was cropped:

{
    "abstract": "\"Single-agent activity for @entity12 reflected by response rates of 10%-30% has been reported in @entity0 with @entity3 ( @entit...",
    "answer": "@entity10",
    "entities_list": ["@entity0", "@entity6", "@entity2", "@entity5", "@entity12", "@entity11", "@entity1", "@entity7", "@entity9", "@entity10", "@entity3", "@entity4", "@entity8"],
    "title": "No synergistic activity of @entity7 and XXXX in the treatment of @entity3 .\n"
}

biomrc_tiny_A

  • Size of downloaded dataset files: 0.02 MB
  • Size of the generated dataset: 0.07 MB
  • Total amount of disk used: 0.09 MB

An example of 'test' looks as follows.

This example was too long and was cropped:

{
    "abstract": "\"OBJECTIVE: Decompressive craniectomy (DC) requires later cranioplasty (CP) in survivors. However, if additional ventriculoperit...",
    "answer": "@entity260 :: (MESH:D011183,Disease) :: ['Postoperative Complications']\n",
    "entities_list": ["@entity1 :: ('9606', 'Species') :: ['Patients', 'patients', 'Patient']", "@entity260 :: ('MESH:D011183', 'Disease') :: ['VPS regarding postoperative complications']", "@entity1276 :: ('MESH:D006849', 'Disease') :: ['hydrocephalus']"],
    "title": "Cranioplasty and Ventriculoperitoneal Shunt Placement after Decompressive Craniectomy: Staged Surgery Is Associated with Fewer XXXX .\n"
}

Data Fields

The data fields are the same among all splits.

biomrc_large_A

  • abstract: a string feature.
  • title: a string feature.
  • entities_list: a list of string features.
  • answer: a string feature.

biomrc_large_B

  • abstract: a string feature.
  • title: a string feature.
  • entities_list: a list of string features.
  • answer: a string feature.

biomrc_small_A

  • abstract: a string feature.
  • title: a string feature.
  • entities_list: a list of string features.
  • answer: a string feature.

biomrc_small_B

  • abstract: a string feature.
  • title: a string feature.
  • entities_list: a list of string features.
  • answer: a string feature.

biomrc_tiny_A

  • abstract: a string feature.
  • title: a string feature.
  • entities_list: a list of string features.
  • answer: a string feature.

Data Splits

biomrc_large_A

train validation test
biomrc_large_A 700000 50000 62707

biomrc_large_B

train validation test
biomrc_large_B 700000 50000 62707

biomrc_small_A

train validation test
biomrc_small_A 87500 6250 6250

biomrc_small_B

train validation test
biomrc_small_B 87500 6250 6250

biomrc_tiny_A

test
biomrc_tiny_A 30

Dataset Creation

Curation Rationale

More Information Needed

Source Data

Initial Data Collection and Normalization

More Information Needed

Who are the source language producers?

More Information Needed

Annotations

Annotation process

More Information Needed

Who are the annotators?

More Information Needed

Personal and Sensitive Information

More Information Needed

Considerations for Using the Data

Social Impact of Dataset

More Information Needed

Discussion of Biases

More Information Needed

Other Known Limitations

More Information Needed

Additional Information

Dataset Curators

More Information Needed

Licensing Information

More Information Needed

Citation Information 

@inproceedings{pappas-etal-2020-biomrc,
    title = "{B}io{MRC}: A Dataset for Biomedical Machine Reading Comprehension",
    author = "Pappas, Dimitris  and
      Stavropoulos, Petros  and
      Androutsopoulos, Ion  and
      McDonald, Ryan",
    booktitle = "Proceedings of the 19th SIGBioMed Workshop on Biomedical Language Processing",
    month = jul,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.bionlp-1.15",
    pages = "140--149",
    abstract = "We introduce BIOMRC, a large-scale cloze-style biomedical MRC dataset. Care was taken to reduce noise, compared to the previous BIOREAD dataset of Pappas et al. (2018). Experiments show that simple heuristics do not perform well on the new dataset and that two neural MRC models that had been tested on BIOREAD perform much better on BIOMRC, indicating that the new dataset is indeed less noisy or at least that its task is more feasible. Non-expert human performance is also higher on the new dataset compared to BIOREAD, and biomedical experts perform even better. We also introduce a new BERT-based MRC model, the best version of which substantially outperforms all other methods tested, reaching or surpassing the accuracy of biomedical experts in some experiments. We make the new dataset available in three different sizes, also releasing our code, and providing a leaderboard.",
}

Contributions

Thanks to @lewtun, @PetrosStav, @lhoestq, @thomwolf for adding this dataset.