--- CORD-19 Dataset --- CORD-19 is a dataset of scientific papers on Covid-19. The papers in the dataset included work on Covid-19, coronaviruses, SARS, MERS, H1N1, and related subjects. Paper metadata are derived and combined from sources such as the WHO, PMC, BioRxiv/MedRxiv, CZI, and Elsevier. Full text of papers is made available where possible by parsing this text out of the accompanying PDF documents. For a count of metadata entries and full text documents each week, please see the accompanying changelog for each release. --- Metadata fields --- Each row of the metadata file corresponds to one paper object. Each unique paper object receives a corresponding `cord_uid`. For each paper, we include references to its other unique identifiers where possible, under the following column headers: - `doi`: Digital Object Identifier (DOI) - `pmcid`: PubMed Central (PMC) identifier - `pubmed_id`: Pubmed Identifier - `Microsoft Academic Paper ID`: Microsoft Academic Graph (MAG) Identifier - `WHO #Covidence`: World Health Organization (WHO) Covidence Identifier Not all papers in this collection have full text. To find the associated full text: - Check column `has_full_text`: if True, there is associated full text - Column `full_text_file` gives the name of the subset file that includes the full text of this paper. This subset file corresponds to the publication license of the paper, available in the `license` column. - Column `sha` gives the associated filenames in the subset file - NOTE: some rows have multiple entries in the `sha` column, separated by ';'. This is due to some papers being associated with multiple document files, e.g. a main paper and several supplementary files. (1) Metadata for papers from these sources are combined: WHO, CZI, PMC, BioRxiv/MedRxiv, Elsevier; see changelog each week for details (2) 17K of the paper records have PDFs and the hash of the PDFs are in 'sha' (3) For PMC sourced papers, one paper's metadata can be associated with one or more PDFs/shas under that paper - a PDF/sha correponding to the main article, and possibly additional PDF/shas corresponding to supporting materials for the article. (4) 13K of the PDFs were processed with fulltext ('has_full_text'=True) (5) Various 'keys' are populated with the metadata: - 'pmcid': populated for all PMC paper records (27337 non null) - 'doi': populated for all BioRxiv/MedRxiv paper records and most of the other records (26357 non null) - 'WHO #Covidence': populated for all CZI records and none of the other records (1236 non null) - 'pubmed_id': populated for some of the records - 'Microsoft Academic Paper ID': populated for some of the records --- Full text files --- Full text is released in the four (4) subset files: - Commercial use subset - Non-commercial use subset - Custom license subset - BioRxiv/MedRxiv subset Documents are separated by copyright license type into these four subsets. Additional licensing information is available on the CORD-19 dataset website: https://pages.semanticscholar.org/coronavirus-research Each of these subsets uncompresses into a directory containing JSON parses of paper full text. To understand how these JSON files are organized, see the JSON schema: https://ai2-semanticscholar-cord-19.s3-us-west-2.amazonaws.com/2020-03-13/json_schema.txt These JSON parses are only meant to represent the full text extracted from the paper documents. The metadata fields in these JSON files may be populated, but we encourage the user to use the metadata given in the metadata file, which have been curated and cleaned. The metadata file provides links between paper entries and full text files (see above).