How do I recover and verify citations from unnamed PDFs left before a lab handoff?

Build a candidate record for every PDF, compare the bibliographic fields against a deposited metadata source, and log any file that cannot be resolved so the handoff ends with verified records, documented duplicates, and a short exception list. The sequence below keeps the work auditable when the original owner is unavailable.

1. Inventory the folder. List every file with its filename, size, and page count, and give each one a working ID (for example pdf-001). Expected result: a fixed list so no file is silently skipped or counted twice.

2. Extract identifiers or title clues. Open each PDF and record any DOI, journal name, volume, page range, or title string visible on the first page or in the header. Expected result: each file either has a DOI to look up or a title clue to search.

3. Create candidate records. For a standalone PDF, Zotero can attempt to create a parent bibliographic item using identifiers or extracted metadata. Expected result: a draft record attached to the file, which you then check and correct rather than accept as final.

4. Compare bibliographic fields. For files with a DOI, look up the DOI through the Crossref REST API, which returns the metadata deposited for that DOI. Compare title, authors, year, journal, volume, and pages field by field against the candidate record. Expected result: a match, a corrected record, or a flagged discrepancy.

5. Resolve duplicates. Run Zotero's duplicate detection, which identifies potential duplicate items and offers a merge interface that preserves collections and tags while combining records. Expected result: one merged record per work, with the merge documented in the handoff log.

6. Assign unresolved files. Any PDF with no identifier, no usable title clue, or a field mismatch that survives comparison goes into the handoff log with its working ID, what was tried, and the specific field still in question. Expected result: a short exception list a colleague can act on without re-reading the whole folder.

7. Close the handoff. Deliver three artifacts: the verified records, the list of documented duplicates and merges, and the exception list. Expected result: the incoming owner knows which citations are ready to use and which still need a decision.

Keep the handoff log alongside the library so later corrections trace back to the file they came from.