Context

In the face of the current disaster, Elsevier opened access to ~20k articles related to COVID-19

Content

I processed all 20k articles shared by Elsevier and selected ~13k those with non-empty titles and abstracts.

• covid_artilces_elsevier_train.csv – 10k processed articles
• covid_artilces_elsevier_validation.csv – 2967 processed articles left for validation (whatever task you come up with)
• meta – a dump of metadata from Elsevier sftp://public@coronacontent.np.elsst.com, fetched on 2020-03-31, 5pm Amsterdam time, XMLs and PDFs are skipped. If you need them, you can download them from the mentioned SFTP server (~5GB XMLs, 20 GB PDFs)