Choose the right source
| Source | Best starting use | Useful identifiers |
|---|---|---|
| Crossref | DOI metadata, publishers, journals, dates, licenses, and reference counts. | DOI |
| OpenAlex | Citation analysis, concepts, institutions, authorships, and open-access fields. | OpenAlex ID, DOI |
| PubMed | Biomedical literature, MeSH-informed searches, publication types, and PMIDs. | PMID, DOI |
Crossref first run
Use Crossref when you need publisher-managed DOI metadata or want to enrich a known DOI list. Open the Crossref Works Scraper and start with:
{
"queries": ["retrieval augmented generation"],
"dois": [],
"maxResults": 25
}Inspect doi, title, publisher, publishedAt, url, referenceCount, and isReferencedByCount. The RAG metadata example is publicly runnable.
OpenAlex first run
Use OpenAlex for broader citation graphs, institutional analysis, and open-access indicators. Open the OpenAlex Works Scraper and use:
{
"queries": ["retrieval augmented generation"],
"dois": [],
"openAlexIds": [],
"maxResults": 25
}Review openAlexId, doi, publicationYear, citedByCount, primaryLocationUrl, and openAccessStatus. Try the public AI citation-data example.
PubMed first run
Use PubMed when the research question is biomedical or clinical. Open the PubMed Articles Scraper and start with:
{
"queries": ["cancer immunotherapy"],
"pmids": [],
"maxResultsPerQuery": 25,
"sort": "relevance"
}The core evidence fields are pmid, title, journal, publicationDate, doi, url, authors, and publicationTypes. The public cancer immunotherapy example provides a bounded starting point.
Check data quality before analysis
- Keep
sourceTypeandsourcein the exported dataset. - Deduplicate by DOI when present. Fall back to the source-specific persistent ID.
- Do not equate a high citation count with study quality.
- Open a sample of source URLs and compare titles, dates, and identifiers.
- Record the run timestamp because citation and indexing metadata changes.
Crossref and OpenAlex currently cost $1.50 per 1,000 returned works; PubMed costs $1.50 per 1,000 returned articles. These are grandfathered acquisition prices. Start with 25 rows even when a large export is inexpensive.
Common questions
Is this the same as scraping Google Scholar?
No. These Actors use Crossref, OpenAlex, and PubMed. They avoid presenting Google Scholar coverage that the workflow does not provide.
Why are citation counts different across sources?
Each source has its own coverage, matching, and update process. Preserve the source label and do not average unlike counts without a documented method.
Can I download the full paper?
These workflows collect public metadata and available links. They do not bypass publisher access controls or guarantee full-text rights.
What should I send to an AI assistant?
Send a bounded set of metadata with persistent IDs and source URLs. Ask the model to separate source facts from its inferences.
Browse all research tools
Compare scholarly metadata Actors and open their public Store pages.
Open academic research tools