Scholarly Metadata: arXiv, OpenAlex and PubMed search

Search three open scholarly sources at once, or look papers up by DOI, arXiv id, PubMed id or OpenAlex id. Every result comes back in one shape, so you can compare and merge them.

For each paper you get:

• Title, authors and abstract
• Publication date, last updated date and year
• DOI, journal or venue, and publication type
• A link to the record, plus the PDF link for arXiv papers and the open access link from OpenAlex
• arXiv categories, OpenAlex citation counts, and the PubMed id where OpenAlex knows it
• The license or attribution line for the source, and for PubMed the publisher's copyright line

Input

• Search queries. Each query runs on every selected source. Words are combined with AND on arXiv. Source syntax passes through: au:Hinton ti:capsule on arXiv, smith j[au] on PubMed.
• Identifiers to look up. The type is detected for you:
  • a DOI (10.1038/nature12373 or a doi.org link) is looked up in OpenAlex and PubMed
  • an arXiv id (1706.03762) or arXiv DOI is looked up in arXiv and OpenAlex
  • a PubMed id (PMID:31452104 or the bare number) is looked up in PubMed and OpenAlex
  • an OpenAlex work id (W2741809807) is looked up in OpenAlex
• Sources: arXiv, OpenAlex, PubMed (default all three).
• Maximum results per query and source: 1 to 200 (default 25).
• Published from and to, as YYYY-MM-DD, for searches.
• Sort searches by relevance or newest first.
• Your own OpenAlex API key and NCBI API key, both optional and stored as secrets.
• A contact email, optional, passed to PubMed and in the User-Agent header.

Output

One dataset item per paper per source, with source, inputType (search or lookup), query or lookupId, recordId, url, title, authors, abstract, publishedDate, updatedDate, year, doi, venue, publicationType, categories, pdfUrl, openAccessUrl, citedByCount, pmid, copyrightNotice, license, attribution and totalResults. A search or lookup that fails or finds nothing gives one item with error, so you can see which input it was.

Authors are listed by name as they appear in the citation. Affiliations and author contact details are not collected.

Source rules and limits

• arXiv: one request every three seconds, one connection at a time, as the arXiv API terms require. arXiv metadata is released under CC0. The actor is not endorsed by arXiv.
• PubMed: 3 requests a second, or 10 with your own NCBI API key. Rows carry the attribution NLM asks for, "Courtesy of the U.S. National Library of Medicine". PubMed abstracts can be under the publisher's copyright: the copyrightNotice field carries the publisher's line when the record has one, and reuse beyond fair use needs the copyright holder's permission.
• OpenAlex: data under CC0. Lookups by id are free. Searches spend a daily budget: without a key it covers about 100 searches a day, and your own free OpenAlex key raises it tenfold. An invalid key gives an error item that says so.

Pricing

Pay per event: one record-returned event per paper returned. Searches and lookups that find nothing or fail are not charged. The run stops when it reaches the maximum charge you set.

Support

Questions and problems: the Issues tab on this page. Replies are written by an AI assistant on behalf of the developer and reviewed by the developer.