Dataset Card for "scifact"

Dataset Summary

SciFact.

This is a dataset of expert-written scientific claims paired with evidence-containing abstracts and annotated with labels and rationales.

Dataset Structure

Data Instances

claims

• Size of downloaded dataset files: 2.72 MB
• Size of the generated dataset: 0.25 MB
• Total amount of disk used: 2.97 MB

An example of 'validation' looks as follows.

{
    "cited_doc_ids": [14717500],
    "claim": "1,000 genomes project enables mapping of genetic sequence variation consisting of rare variants with larger penetrance effects than common variants.",
    "evidence_doc_id": "14717500",
    "evidence_label": "SUPPORT",
    "evidence_sentences": [2, 5],
    "id": 3
}

corpus

• Size of downloaded dataset files: 2.72 MB
• Size of the generated dataset: 7.63 MB
• Total amount of disk used: 10.35 MB

An example of 'train' looks as follows.

This example was too long and was cropped:

{
    "abstract": "[\"Alterations of the architecture of cerebral white matter in the developing human brain can affect cortical development and res...",
    "doc_id": 4983,
    "structured": false,
    "title": "Microstructural development of human newborn cerebral white matter assessed in vivo by diffusion tensor magnetic resonance imaging."
}

Data Fields

The data fields are the same among all splits.

claims

• id: a int32 feature.
• claim: a string feature.
• evidence_doc_id: a string feature.
• evidence_label: a string feature.
• evidence_sentences: a list of int32 features.
• cited_doc_ids: a list of int32 features.

corpus

• doc_id: a int32 feature.
• title: a string feature.
• abstract: a list of string features.
• structured: a bool feature.

Data Splits

claims

	train	validation	test
claims	1261	450	300

corpus

	train
corpus	5183

Additional Information

Licensing Information

https://github.com/allenai/scifact/blob/master/LICENSE.md

The SciFact dataset is released under the CC BY-NC 2.0. By using the SciFact data, you are agreeing to its usage terms.

Citation Information

@inproceedings{wadden-etal-2020-fact,
    title = "Fact or Fiction: Verifying Scientific Claims",
    author = "Wadden, David  and
      Lin, Shanchuan  and
      Lo, Kyle  and
      Wang, Lucy Lu  and
      van Zuylen, Madeleine  and
      Cohan, Arman  and
      Hajishirzi, Hannaneh",
    booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
    month = nov,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2020.emnlp-main.609",
    doi = "10.18653/v1/2020.emnlp-main.609",
    pages = "7534--7550",
}