Near-duplicate detection measures textual similarity between documents and groups those exceeding a threshold. Unlike hash-based deduplication, it identifies documents that differ in minor ways such as formatting, dates, or small edits.
Similarity thresholds are configurable.
Alternative Names:
Near-Dupe Detection, Similarity Detection
Why it Matters?
Grouping near-duplicates improves review consistency, since a reviewer seeing eight drafts of the same contract together codes them coherently rather than encountering them scattered across a review population. It also surfaces the differences between versions, which is frequently the evidentiary point: what changed between drafts and when. Threshold setting matters, since a loose threshold groups unrelated documents.
Frequently Confused with
Related terms
Frequently asked questions
How does near-duplicate detection differ from deduplication?
Why group near-duplicates for review?
