Comparing https://about.lens.org/covid-19/ scholary works and CORD19 Dataset?

[motey]

At the moment we are using the CORD19 Dataset for importing scientific papers to the covidgraph.
https://www.kaggle.com/allen-institute-for-ai/CORD-19-research-challenge

I just stumbled over a similar dataset at https://about.lens.org/covid-19/ "Scholary Works"

It would be interesting to know if this dataset is of similar scope and better quality and can maybe replace the CORD19 dataset. As the CORD19 Dataset is of poor quality in many places.

Task: Create a comparison between the CORD19 and lens.org Scholary Works dataset:

• Scope / Amount of articles?

• Which dataset comes with more identifieng attributes like DOIs, PCMiD, PubMedIDs? In which dataset these are more consistently appearing on each article?

• Which dataset has more relevant attributes (e.g. MeshTerms) ?

• Which Dataset is better for distincting authors (e.g. brings ORCID for some authors, etc)?

Extra Task: Find a even better datasource :)

[motey]

Pedro Parraguez: we have use extensibly the Microsoft Academic Graph dataset that for the most part is what Lens uses for Scholarly works. It is indeed of quite good quality. You don´t get the full text but in return you get better metadata, less duplicates, etc. As far as I have seen the coverage of pre-prints is lower in Lens, though. Maybe DOI-level matching would allow having the best of both worlds (although gaps will exist).