How to read a Hallucite result and its evidence
By Hallucite · Content reviewed
Run a check, then read the evidence behind each result. This guide describes the current reference comparison interface and its publication-check scope. Start with a row, open its comparison details and inspect what was cited, what record was found and which checks answered. Keep the methodology and limitations beside your interpretation: a verdict is about publication facts, not authorship or claim support.
Start with the row verdict
The current Hallucite check uses row verdicts including Verified, Discrepancy, Not Found, Could not verify and Could not parse. Pending and Error describe unfinished or failed work. Read the row label together with its details. Choose the entry you want to investigate and open its comparison details before deciding whether to edit the bibliography.
Verified reports a publication match within the evidence checked. Discrepancy directs attention to differences, while Not Found leaves an entry to investigate. Could not verify and Could not parse require attention to checking or input limits. Hallucite’s limitations summary explains that missing evidence does not establish fabrication. Preserve the specific issue instead of treating every non-matching result as the same problem.
Compare the cited fields with the found record
The current comparison interface shows “Parsed Citation Fields” when parsed citation data is available; otherwise it shows the raw reference text. Check that the displayed input information corresponds to the entry you meant to submit before interpreting a difference. If extracted fields do not reflect the source text, review the input before making a publication correction.
When shown, “Matched Information” lists available title, journal, year and DOI details. The current result view links a matched title or DOI when a usable source link exists. Open that record and compare it with the intended work. Record which field differs; do not replace the entire reference simply because a candidate record was found.
Read “Field comparison” and “Checks run”
Where an evidence summary is available, the current evidence panel shows a headline and a “Checks run” section. When field evidence is present, “Field comparison” appears above it with the columns “Field”, “Status”, “You cited” and “We found”. These sections render expanded in comparison details. Read the relevant field beside the verdict rather than copying only the badge into your notes. Keep the underlying record link with any correction you make.
The field comparison uses states such as matches, close, differs, not cited and not in record in the current interface. A missing field and a conflicting field need different investigations. If the reference omits a value, consult the source before adding one. If the record omits a value, look at the publication itself rather than assuming your cited value is wrong.
Inspect which sources answered
The current details and evidence panels show available source outcomes and coverage caveats. Read an unavailable or skipped check as a limitation of that check, rather than as evidence that the publication does not exist. For an unresolved reference, note the evidence actually returned and the next suitable catalogue or publisher search you can perform.
Hallucite’s methodology names source outages, incomplete metadata and coverage gaps as limits. These matter when interpreting a missing match. If a source was unavailable, retry later or look directly at an appropriate publication record. Keep “no match in the evidence checked” separate from a claim that every possible source has been searched.
Worked example: turn a field difference into a review note
This illustrative manual comparison is not a captured checker result. Deliberately altered input: Turing, A. M. (2020). Computing machinery and intelligence. Mind, 59(236), 433–460. DOI: 10.1093/mind/LIX.236.433. The Oxford Academic publisher record instead identifies Turing, A. M. (1950). Computing machinery and intelligence. Mind, 59(236), 433–460. at that same DOI.
The comparison to record is specific: cited year 2020; publisher year 1950. A useful review note is “year differs from publisher record; confirm the intended work and correct the year”. Keep the record link with the note instead of turning the difference into a finding about authorship.
Next action: inspect the original publication, correct the entry if it is the intended work, and read the relevant passage to assess claim support. We have deliberately changed one field to explain evidence review; no live verdict, accuracy measurement or publication-status assessment is asserted.
Inspect retraction information separately
In the current results table, a retracted result leads with Retracted, while record identity can be shown separately. Open the comparison details and review available retraction information. Establish which publication the notice concerns and read the notice itself before changing how the source is used. A publication match does not remove a status warning.
For independent context, Crossref’s Retraction Watch documentation describes retractions and other updates, and NLM’s policy explains linked notices in PubMed. Keep the original work and the notice as separate records. If the retraction check is inconclusive, preserve that limitation and investigate a source central to your argument directly.
Save the evidence and choose the next action
When source assessments are present, the current comparison details offer a Verification evidence section with structured identity, coverage, source, retraction and evidence data. Use it to inspect the available information. For a practical review note, save the reference, the field or status needing attention, the source link and your decision about the correction or further search.
After correcting an entry, inspect its citation in the manuscript and read the supporting passage. Hallucite’s methodology keeps publication verification separate from claim support and authorship. Finish with a corrected bibliography and explicit unresolved questions for your co-authors. Review sources added during revision in the same way, keeping the evidence behind each decision instead of treating the result list as a blanket guarantee.
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