# Making Science Simple: Corpora for the Lay Summarisation of Scientific Literature

**Authors:** Tomas Goldsack, Zhihao Zhang, Chenghua Lin, Carolina Scarton  
**arXiv:** [2210.09932](https://arxiv.org/abs/2210.09932)  
**Date:** 2022-10-18  
**Summary page:** https://www.summarizepaper.com/en/paper/2210.09932v2/

## Key points

- Lay summarisation is important for making scientific texts more understandable for non-experts.
- Automatic approaches are needed to broaden access to scientific literature and facilitate interdisciplinary knowledge sharing.
- Current corpora for lay summarisation are limited, hindering effective data-driven approaches.
- Two new datasets have been introduced: PLOS (large-scale) and eLife (medium-scale), containing biomedical journal articles paired with expert-written lay summaries.
- The researchers thoroughly analyzed these summaries and benchmarked them using mainstream summarisation approaches.
- A manual evaluation with domain experts was conducted to demonstrate the utility of the lay summaries and highlight key challenges associated with lay summarisation.
- The code and datasets are available through a GitHub repository.
- Background information on PLOS as an open-access publisher hosting influential peer-reviewed journals across various scientific fields is provided, along with details about eLife as an open-access journal focusing on biomedical and life sciences.
- Analyses comparing technical abstracts with lay summaries from both datasets were conducted to quantify differences in readability, rhetorical structure, vocabulary sharing, and abstractiveness.
- The study contributes valuable insights into improving lay summarisation techniques by providing comprehensive datasets for further research in this area.

## Summary

In this study, the researchers focus on lay summarisation and its importance in making scientific texts more understandable for non-experts. They highlight the need for automatic approaches in broadening access to scientific literature and facilitating interdisciplinary knowledge sharing. However, current corpora for this task are limited, hindering effective data-driven approaches. To address this issue, the researchers introduce two new datasets: PLOS (large-scale) and eLife (medium-scale), containing biomedical journal articles paired with expert-written lay summaries. The researchers thoroughly analyze these summaries and benchmark them using mainstream summarisation approaches. They also conduct a manual evaluation with domain experts to demonstrate their utility and shed light on key challenges associated with lay summarisation. The code and datasets are available through a GitHub repository. Additionally, background information on PLOS as an open-access publisher hosting influential peer-reviewed journals across various scientific fields is provided. Similarly, eLife is an open-access journal focusing on biomedical and life sciences where selected articles receive simplified summaries written by expert editors. The study includes analyses comparing technical abstracts with lay summaries from both datasets to quantify differences in readability, rhetorical structure, vocabulary sharing, and abstractiveness. Overall, this study contributes valuable insights into improving lay summarisation techniques by providing comprehensive datasets for further research in this area.