← allmeta

Citation Dedup

Given a list of citations, flag likely duplicates via DOI match, normalized title similarity (trigram Jaccard), first-author surname, and year. Export the cleaned set as CSV. Everything runs in your browser.

Input

Paste citations Format Citations CSV: first line headers optional. Case and punctuation in titles are normalized.

Matching Title similarity threshold (Jaccard) Lower = more aggressive (more groups). 0.85 ≈ typical SR dedup. Year tolerance (± years) Require first-author surname match

Results

Input records0

Unique clusters0

Duplicates flagged0

Max cluster size0

Cluster-size histogram

Title normalisation: lowercase, strip punctuation, collapse whitespace, drop stopwords (a/an/the/of/in/on/for/and/or/to). Similarity: Jaccard on character trigrams (n=3) over the normalised title. Behaviourally verified: DOI-variant merging, title near-duplicate merging, distinct-paper separation, the year-tolerance and same-author guards, and the normalisation/Jaccard primitives are pinned by deterministic constructed-oracle tests (hub/shared/tests/citation-dedup-behavior.spec.mjs).