Crossref Scholarly Works Scraper

Search the Crossref catalog of 150M+ scholarly works (journal articles, preprints, books, datasets, and more) via its public REST API — no API key, no login, no anti-bot.

The actor is a polite Crossref client: it identifies itself with a contact User-Agent and a mailto query parameter so Crossref routes it to the faster "polite pool", and it uses deep cursor pagination (cursor=* → next-cursor) which is the only reliable way to page past 1,000 rows.

Input

Field	Type	Default	Description
query	string (required)	deep learning	Keywords searched across titles, authors, abstracts and metadata.
filterType	string	all	Restrict to a Crossref work type, e.g. journal-article.
fromDate	string YYYY-MM-DD	none	Only works published on/after this date.
sort	enum	relevance	relevance, is-referenced-by-count (most cited), or published (newest).
maxItems	integer	100	Max works to return (cursor pagination handles >100).
proxyConfiguration	object	none	Optional and off by default; Crossref is a public, no-key API with no anti-bot, so a proxy adds no benefit. Only enable it if you hit IP-level rate limits.

Output

Each successful row:

{
  "ok": true,
  "doi": "10.1038/nature14539",
  "title": "Deep learning",
  "authors": ["Yann LeCun", "Yoshua Bengio", "Geoffrey Hinton"],
  "journal": "Nature",
  "publisher": "Springer Science and Business Media LLC",
  "type": "journal-article",
  "publishedDate": "2015-05-28",
  "citations": 70000,
  "subjects": ["Multidisciplinary"],
  "issn": ["0028-0836", "1476-4687"],
  "abstract": null,
  "url": "https://doi.org/10.1038/nature14539"
}

• authors are formatted "Given Family" (organizational authors fall back to their name).
• publishedDate is assembled from Crossref's date-parts (may be year-only or year-month for older records).
• citations is Crossref's is-referenced-by-count.
• abstract is the JATS-XML abstract stripped to plain text, or null when Crossref has none.
• Nullable fields: title, journal, publisher, type, publishedDate, abstract, and url may be null, and authors, subjects, and issn may be empty arrays, depending on what the publisher deposited with Crossref. doi is always present (rows without a DOI are dropped). citations defaults to 0 when absent.

Results are deduplicated by DOI. Charging is per successful work (work event). Diagnostic / empty / blocked rows (ok: false with an errorCode) are never charged — this includes BAD_INPUT (empty query or malformed fromDate), NO_RESULTS, and any network/block error.

Troubleshooting

• BAD_INPUT row, no results: you left query empty or fromDate isn't YYYY-MM-DD. Fix the input and re-run — you were not charged.
• NO_RESULTS row: your query/filter combination matched nothing in Crossref. Try broader keywords or drop the type/date filters.
• RATE_LIMITED / BLOCKED row: rare for Crossref. The actor already retries with backoff; if it persists, enable a proxy to use a different IP.

Notes

• Powered entirely by the public Crossref REST API (https://api.crossref.org/works). Please be considerate of the shared, free service.
• Citation counts and abstracts depend on what publishers deposit with Crossref; coverage varies by record.