Research Paper Scraper — Citations & Retractions

Scrape academic papers — titles, abstracts, authors, institutions, citations — from OpenAlex, an open catalogue of over 250 million works.

The Google Scholar scrapers on this store are rated around 2.5 stars, and not because they are badly written. Scholar blocks automated traffic, so they fail. OpenAlex has a public API with no key, no bot wall and no CAPTCHA. Same literature, a source that cannot lock you out. That is the whole competitive argument, and it is worth more than any feature.

Three fields Scholar will never give you

1. The abstract, rebuilt. OpenAlex is not permitted to redistribute abstracts as prose, so it ships an inverted index —

{"gene": [3], "the": [0, 12]}

2. Field-weighted citation impact (fwci). Forty citations is outstanding in one field and invisible in another, so a raw count cannot be compared across disciplines. FWCI normalises against the work's own field and year: 1.0 is exactly average, 2.0 is twice the field average. 196 of 200 papers in that same live query carried one. Also included is citationPercentileMin/Max — the percentile band within field and year.

3. Retraction flags. OpenAlex marks 134,038 retracted works (measured 2026-08-14). The live CRISPR query surfaced 2 in its top 200. A literature review or an AI pipeline that cites a retracted paper has a real problem, and Scholar will not warn you.

What else you get

• Collaboration structure. Author institutions with ROR ids, distinct countries, and isInternationalCollaboration.
• Open access. isOpenAccess, openAccessStatus (gold, green, hybrid, bronze) and a direct link to the legal full text.
• An analysis row over the whole query: output by year, median FWCI, how many works beat their field average, top-1% count, uncited share, retraction rate, open-access share, and the top institutions, authors, journals and countries.

A live run on CRISPR, 2024 onward, 200 most-cited papers: median FWCI 19.9, 196 above field average, 122 in the top 1% of their field, 75.5% open access, 44% international. Top institutions Harvard (13), UC Berkeley (13), Broad Institute (12); top journals Nature (18), Nature Communications (14), Cell (12).

Who uses this

• Biotech and pharma competitive intelligence — who publishes in a field, where, and with what real impact.
• Research offices and funders — benchmarking an institution's output and citation performance.
• Systematic reviewers — bulk retrieval with retraction screening built in.
• AI and RAG pipelines — abstracts and metadata from a source that will not rate-limit a crawl into failure.

Input

Give at least one narrowing filter — searchTerm, institutionRor, countryCode, authorId, or retractedOnly. An unfiltered walk across 250 million works is refused rather than started.

Output

work — title, doi, publicationDate, workType, citedByCount, fwci, citationPercentileMin/Max, isRetracted, isOpenAccess, openAccessUrl, sourceName, publisher, primaryTopic/Field/Domain, authors, firstAuthor, correspondingAuthors, institutions, institutionRors, countries, isInternationalCollaboration, abstract.

analysis — worksByYear, medianCitations, medianFwci, worksAboveFieldAverage, top1PercentWorks, uncitedSharePercent, retractedWorks, openAccessSharePercent, internationalCollaborationSharePercent, topInstitutions, topAuthors, topSources, topTopics, topCountries.

Notes on accuracy

• top1PercentWorks reads the lower bound of the percentile band. A work in the 94–100 band is not counted, because an overlapping band is not evidence. Overstating impact is the failure mode that matters in this dataset.
• Not every work has an FWCI — very new and very obscure works may not.
• Abstracts are missing where OpenAlex holds no inverted index for the work.
• The actor identifies itself to OpenAlex's "polite pool", which is faster and more reliable than anonymous access.

Pricing

Pay per event. $0.001 per paper, $0.05 per analysis row.