Read every common format
Paste formatted references or upload PDF, DOCX, BibTeX, RIS, EndNote, NBIB, XML, JSON, CSV, or TSV.
Check a reference list against scholarly sources, find missing or incorrect fields, and spot citations that may have been fabricated by AI.
Paste citations below or upload a bibliography file. LumaCite separates the records and checks them one by one.
Ready. Paste citations or choose a supported file.
The evidence report will appear here.
Paste formatted references or upload PDF, DOCX, BibTeX, RIS, EndNote, NBIB, XML, JSON, CSV, or TSV.
Recover clean citation rows and detect DOI, PMID, PMCID, ISBN, arXiv, ISSN, and article URLs.
Use identifiers first, then title, author, year, and venue matching when a stable identifier is missing.
Compare publisher, biomedical, book, data, preprint, and scholarly-graph metadata sources.
Show matches, missing values, conflicting values, enrichments, duplicates, and source-level provenance.
Separate verified citations from fixable errors and unverified records with possible fabrication risk.
Exact-identifier and fuzzy bibliographic matching with explainable confidence.
Input-versus-source comparisons for title, authors, year, venue, volume, issue, pages, and IDs.
Retraction, correction, Crossmark, open-access, and registry availability signals.
Readable reports for humans plus downloadable JSON evidence for reproducible review.
The checker first resolves a stable identifier when one is present. It then compares the supplied title, authors, date, publication, volume, issue, pages, and identifiers with independent metadata records. A high score requires identity evidence, not merely complete-looking text.
No checker can prove authorship from a citation string alone. LumaCite reports possible fabrication risk when a reference cannot be corroborated, or when its identifier resolves to a different work. It shows the failed and conflicting checks so a person can make the final decision.
No. Many valid books, older articles, websites, theses, and reports do not have a DOI. Missing identifiers lower automated confidence, but LumaCite also searches by title, author, year, venue, ISBN, PMID, and arXiv identifiers when relevant.
The checker accepts PDF, DOCX, TXT, Markdown, RTF, BibTeX, RIS, EndNote tagged text, PubMed NBIB, EndNote XML, CSL-JSON, CSV, and TSV. Running a check does not add the citations to your LumaCite library.