Dedupe and merge

The problem

The same paper got imported three times — once from the browser connector, once from a BibTeX file, once by DOI — with slightly different titles each time. Zotero’s built-in duplicate detection is conservative and its merge is manual, one group at a time. You want to find near-duplicates across the whole library and clear them out.

Via an AI agent

“Find duplicate titles in collection Reading list, including fuzzy matches, show me the groups, then merge each group keeping the oldest as the master. Back up first.”

The agent takes a backup, finds duplicate groups (exact or fuzzy), shows them for your approval, and merges — oldest item as master.

Via the CLI

zot dedupe --by title --collection SS5MVVB6           # exact-title duplicates
zot dedupe --by title --fuzzy --collection SS5MVVB6   # near-duplicates (Levenshtein)
zot dedupe --by doi                                    # duplicates by DOI

Review the groups, then merge (oldest item becomes the master):

zot backup                          # REQUIRED — merges cannot be undone
zot dedupe --by title --merge --yes

Scope with --collection before ever running library-wide.

Safety net

Merges are irreversible

Unlike zot apply and zot enrich, a merge cannot be undone by zot undo. The only recovery is the database backup. Always run zot backup immediately before zot dedupe --merge, and disable auto-sync so a bad merge doesn’t propagate to zotero.org.

zot backup                                   # snapshot zotero.sqlite
zot dedupe --by title --fuzzy --dry-run      # inspect groups before committing
# ... verify the groups are true duplicates ...
zot dedupe --by title --merge --yes

If a merge went wrong, restore the zotero.sqlite snapshot that zot backup printed the path to.