Dataset Card for "scientific_lay_summarisation"

• Repository: https://github.com/TGoldsack1/Corpora_for_Lay_Summarisation
• Paper: Making Science Simple: Corpora for the Lay Summarisation of Scientific Literature
• Size of downloaded dataset files: 850.44 MB
• Size of the generated dataset: 1.32 GB
• Total amount of disk used: 2.17 GB

Dataset Summary

This repository contains the PLOS and eLife datasets, introduced in the EMNLP 2022 paper "Making Science Simple: Corpora for the Lay Summarisation of Scientific Literature " .

Each dataset contains full biomedical research articles paired with expert-written lay summaries (i.e., non-technical summaries). PLOS articles are derived from various journals published by the Public Library of Science (PLOS), whereas eLife articles are derived from the eLife journal. More details/analyses on the content of each dataset are provided in the paper.

Both "elife" and "plos" have 6 features:

- "article": the body of the document (including the abstract), sections separated by "/n".
- "section_headings": the title of each section, separated by "/n". 
- "keywords": keywords describing the topic of the article, separated by "/n".
- "title": the title of the article.
- "year": the year the article was published.
- "summary": the lay summary of the document.

Note: The format of both datasets differs from that used in the original repository (given above) in order to make them compatible with the run_summarization.py script of Transformers. Specifically, sentence tokenization is removed via " ".join(text), and the abstract and article sections, previously lists of sentences, are combined into a single string feature ("article") with each section separated by "\n". For the sentence-tokenized version of the dataset, please use the original git repository.

Supported Tasks and Leaderboards

Papers with code - PLOS and eLife.

Languages

English

Dataset Structure

Data Instances

plos

• Size of downloaded dataset files: 425.22 MB
• Size of the generated dataset: 1.05 GB
• Total amount of disk used: 1.47 GB

An example of 'train' looks as follows.

This example was too long and was cropped:
{
    "summary": "In the kidney , structures known as nephrons are responsible for collecting metabolic waste . Nephrons are composed of a ...",
    "article": "Kidney function depends on the nephron , which comprises a 'blood filter , a tubule that is subdivided into functionally ...",
    "section_headings": "Abstract\nIntroduction\nResults\nDiscussion\nMaterials and Methods'",
    "keywords": "developmental biology\ndanio (zebrafish)\nvertebrates\nteleost fishes\nnephrology",
    "title": "The cdx Genes and Retinoic Acid Control the Positioning and Segmentation of the Zebrafish Pronephros",
    "year": "2007"
}

elife

• Size of downloaded dataset files: 425.22 MB
• Size of the generated dataset: 275.99 MB
• Total amount of disk used: 1.47 MB

An example of 'train' looks as follows.

This example was too long and was cropped:
{
    "summary": "In the USA , more deaths happen in the winter than the summer . But when deaths occur varies greatly by sex , age , cause of ...",
    "article": "In temperate climates , winter deaths exceed summer ones . However , there is limited information on the timing and the ...",
    "section_headings": "Abstract\nIntroduction\nResults\nDiscussion\nMaterials and methods",
    "keywords": "epidemiology and global health",
    "title": "National and regional seasonal dynamics of all-cause and cause-specific mortality in the USA from 1980 to 2016",
    "year": "2018"
}

Data Fields

The data fields are the same among all splits.

plos

• article: a string feature.
• section_headings: a string feature.
• keywords: a string feature.
• title : a string feature.
• year : a string feature.
• summary: a string feature.

elife

• article: a string feature.
• section_headings: a string feature.
• keywords: a string feature.
• title : a string feature.
• year : a string feature.
• summary: a string feature.

Data Splits

name	train	validation	test
plos	24773	1376	1376
elife	4346	241	241

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

"Making Science Simple: Corpora for the Lay Summarisation of Scientific Literature"
Tomas Goldsack, Zhihao Zhang, Chenghua Lin, Carolina Scarton
EMNLP 2022