How to Search Crossref DOI Metadata

Crossref provides a rich public metadata service, but repeatable discovery still needs query storage, pagination, normalization, and export. The Crossref DOI Scraper turns text searches into structured publication rows.

Define the search population

Start with one topic and a small result cap. Add date or type filters only when they reflect the research question.

{
  "queries": ["agentic artificial intelligence"],
  "fromPublishedDate": "2024-01-01",
  "workType": "journal-article",
  "maxResultsPerQuery": 50,
  "mailto": "research@example.org"
}

Rows include DOI, titles, type, publisher, container title, authors and affiliations, several publication dates, Crossref timestamps, cited-by and reference counts, ISSNs, ISBNs, language, licenses, and the canonical resource URL.

Inspect member-deposited metadata

Open DOI links for important works and compare title, authors, dates, and venue with the publisher. Crossref stores metadata deposited by members, so missing values and formatting differences are normal.

Use normalized DOI as the primary key. Do not deduplicate on title because editions, corrections, chapters, and distinct works can share similar text. Keep the query and collection time with every row.

If the returned count reaches the configured cap, label the search truncated or narrow it. Citation indicators are Crossref-specific and should not be presented as universal totals. For recurring discovery, save the input as an Apify Task and compare only successful datasets. A work newly observed in a run may be newly deposited or newly matched, not newly published; use published and created dates to distinguish those cases.