For editors and reviewers: checking reference flags
By Hallucite · Content reviewed
Check publication facts before drawing conclusions about a submission. Use reference screening to focus review and return specific, checkable findings to authors. A flag is a prompt to inspect evidence, not an authorship finding; see Hallucite’s limitations.
Signals worth a second look
- A reference that does not resolve: check the title and identifier separately, then broaden the search.
- A DOI that reaches a different paper: compare the destination with the cited work.
- An author list that differs from the record: inspect the full publication before requesting a correction.
Use the DOI resolver and publication metadata for comparison. Formatting and unfamiliar names alone are not grounds for a finding.
A lightweight screening workflow
- Run the reference list through an existence and consistency check.
- Review flags and the sources supporting important claims.
- For each flag, inspect the record, identifier and search limitations.
- Return the precise discrepancy and supporting link to the author.
Follow the venue’s own process. For example, the ICLR policy statement discusses reference issues and conference decisions; an automated flag is not a substitute for that process.
Be fair about unresolved results
Not every unverifiable reference is fake. Hallucite’s methodology describes coverage gaps, incomplete metadata and source outages. Investigate books, proceedings and non-English works with an appropriate library or publisher search. Keep a factual record of what you could establish.
Screening at journal scale
Hallucite returns per-reference outcomes and available evidence for review. See how verification works and the research page for measurement context. Public benchmark publication is pending review; this guide makes no accuracy claim.
Check a reference list against publication records
Hallucite compares references with available publication records and shows the evidence. Review flags and unresolved entries; a result does not establish authorship or claim support. Read the methodology and limitations