Crossref DOI Metadata Search & Enrichment

Resolve a list of Crossref DOIs or search scholarly works by topic. Get flat publication metadata ready for CSV, JSON, spreadsheets and reference-enrichment pipelines. No API key or login required.

The Actor uses Crossref's documented public REST API. It does not fetch publisher websites, PDFs, article bodies, abstracts, author profiles or contact information. Person-level fields and abstracts are never written to its dataset. Lookup responses are normalized in memory and discarded; search requests select only the output metadata fields.

Pricing

$1 per 1,000 successfully resolved records. $0.001 per run. Standard-run platform usage is included. Only successful eligible rows in the default dataset trigger the result charge. Error, not-found, skipped and robots-denied diagnostics do not trigger a result charge. The run fee, where listed, still applies to empty or failed runs. Memory is limited to 256–512 MB. Set a maximum run cost in Apify; inspect the actual dataset and run charge breakdown after an interruption. Input item limits bound work and do not replace the run spending limit.

What you get

• DOI and resolver link, title, publication type, publisher and journal/container title.
• Publication date with its original precision: year, month or day.
• Crossref-reported reference and citation counts, with missing values kept null.
• Deposited license URLs and update relations as evidence, not unsupported legal or research-integrity conclusions.
• Source endpoint and retrieval timestamp on every row.
• A separate SUMMARY record describing limits, unavailable data and unknown DOIs.

Quick start

Look up a DOI:

{"mode":"lookup","dois":["10.1038/nphys1170"],"maxItems":10}

Search a topic:

{"mode":"search","query":"solar cell efficiency","fromDate":"2024-01-01","workType":"journal-article","sort":"published","maxItems":100}

In Apify, open the dataset to export JSON, CSV or Excel. The output tab links to both the dataset and SUMMARY.

Input

Field	Default	Meaning
mode	lookup	lookup or search
dois	none	1–100 Crossref DOIs or doi.org URLs; lookup only
query	none	Required keyword phrase in search mode, maximum 300 characters
maxItems	100	Global unique-output cap, 1–1,000
pageSize	100	Search rows per request, 1–1,000; constant during pagination, initially capped at maxItems
maxPages	20	Search page cap, 1–50
fromDate / toDate	none	Search publication-date bounds in YYYY-MM-DD
workType	all	Optional supported Crossref publication type
sort	relevance	relevance, published, is-referenced-by-count or updated; descending

Search and lookup fields cannot be combined. Exact duplicate DOIs are normalized and requested once per run. No email is transmitted to Crossref and the Actor uses the public pool. A descriptive User-Agent identifies the software and its public developer page. All requests are sequential with at least 1.1 seconds between starts. This meets the July 2026 public-list limit of one request per second; stricter response headers and Retry-After delays are respected.

Output semantics

Each dataset row represents one unique DOI. See dataset-item.schema.json for the full contract.

• publicationDate can be 2020, 2020-02, 2020-02-29 or null. datePrecision tells you which.
• citedByCount is Crossref's count, not a complete citation census or an impact factor. Missing is null; a reported zero stays zero.
• licenses are deposited links, not a verified determination that a publication is open access or that reuse is lawful. Content version and start date matter.
• updates contains Crossref's update-to relationships. An empty list does not prove a work has never been corrected or retracted. No boolean safety/retraction certificate is inferred.
• sourceUrl identifies the official record API. doiUrl points to the DOI resolver and is never automatically fetched.

SUMMARY status can be complete, no_matches, limited, partial or error. notFound lists requested DOIs absent from Crossref. A 404 is not proof that the DOI is invalid: another registration agency may manage it. An all-error run fails after writing SUMMARY, instead of silently succeeding with an empty dataset.

Search uses documented offset pagination within the first 10,000 source positions. Rows stay constant during a run, while dataset output is trimmed to maxItems. It stops on a short page, maximum items, maximum pages, no new DOIs, or the source-position bound. Duplicate/non-progress pages and malformed rows are disclosed. An output cap is not a claim of exhaustive retrieval. Pagination is not a frozen snapshot; the upstream index can change, causing omissions or duplicates. DOI deduplication removes repeats but cannot recover omitted records. The Actor does not offer bulk corpus harvesting or cursor pagination.

Limits and responsible use

• Crossref metadata coverage depends on deposits and updates. Missing or outdated records are possible.
• No full text, abstracts, emails, author names, ORCID profiles, institution contacts, paywall access or non-Crossref enrichment.
• No browser, residential proxy, rotating IP, CAPTCHA handling or access-control bypass. HTTP 403 stops further requests in the run. Transient errors receive at most three attempts; an exhausted rate limit or source failure stops subsequent DOI requests too.
• Responses are capped at 5 MB and requests at 30 seconds. A long Retry-After interval stops the request with a clear rerun-later status rather than retrying early.
• Do not enter sensitive or secret information in queries. Queries and DOIs go to Crossref. The platform also stores the task input and summary in the operator's account.
• This is independent software, not affiliated with or endorsed by Crossref.

License and source documentation

Crossref's metadata licensing statement says bibliographic facts can be reused without restriction and Crossref-generated data is CC0. Abstracts retain publisher or author copyright, so this Actor excludes them. A publication's deposited content license is separate from the metadata license.

• Official REST API
• Access and authentication
• July 2026 rate-limit update
• Official offset pagination and parameter documentation
• August 2026 cursor restrictions