Zenodo Open Metadata snapshot - Training dataset for records and communities classifier building

<p>This dataset contains Zenodo&#39;s published open access records&nbsp;and communities&nbsp;metadata, including entries marked by the Zenodo staff as spam and deleted.</p> <p>The datasets are&nbsp;gzipped compressed&nbsp;JSON-lines&nbsp;files, where each line is a JSON object representation of a Zenodo record or community.</p> <p><strong>Records dataset</strong></p> <p>Filename:<strong> </strong>zenodo_open_metadata_{ date of export }.jsonl.gz</p> <p>Each object&nbsp;contains the terms:&nbsp;<em>part_of,&nbsp;thesis, description, doi, meeting, imprint, references, recid, alternate_identifiers, resource_type, journal, related_identifiers,&nbsp;title, subjects, notes, creators, communities, access_right,&nbsp;keywords, contributors, publication_date</em></p> <p>which correspond&nbsp;to the fields with the same name available&nbsp;in Zenodo&#39;s record JSON Schema at&nbsp;<a href="https://zenodo.org/schemas/records/record-v1.0.0.json">https://zenodo.org/schemas/records/record-v1.0.0.json</a>.</p> <p>In addition, some terms have been altered:</p> <ul> <li>The term <strong>files</strong>&nbsp;contains a list of dictionaries containing <strong>filetype</strong>, <strong>size,</strong>&nbsp;and <strong>filename&nbsp;</strong>only.</li> <li>The term <strong>license</strong>&nbsp;contains a short Zenodo ID of the license (e.g.&nbsp;&quot;cc-by&quot;).</li> </ul> <p><strong>Communities dataset</strong></p> <p>Filename:<strong> </strong>zenodo_community_metadata_{ date of export }.jsonl.gz</p> <p>Each object&nbsp;contains the terms: <em>id, title, description, curation_policy, page&nbsp;</em></p> <p>which&nbsp;correspond&nbsp;to the fields with the same name available&nbsp;in Zenodo&#39;s community creation form.</p> <p><strong>Notes for all&nbsp;datasets</strong></p> <p>For each object the term <strong>spam</strong>&nbsp;contains a boolean value, determining whether a given record/community was marked as&nbsp;spam content&nbsp;by Zenodo staff.</p> <p>Some values for the top-level terms, which were missing in the metadata may contain a&nbsp;<strong>null</strong> value.</p> <p>A smaller uncompressed random sample of 200 JSON lines is&nbsp;also included for each dataset&nbsp;to test and get familiar with the format without having to download the entire dataset.</p>