Run it in your browser

Upload a ticket CSV export with id, requester, subject, body, created_at and status columns, and download the exact files the CLI writes: clusters.csv and repeat-contacts.md. Review only, nothing is merged or changed.

Tickets CSV file

Similarity threshold (0 to 1)

CLI flag: --threshold. Default 0.8.

Pairing window (days)

CLI flag: --window-days. Default 7, boundary inclusive.

Repeat contact window (days)

CLI flag: --repeat-days. Default 30. Used by the repeat contacts report.

Cluster	Ticket IDs	Reason	Confidence

repeat-contacts.md

What it does

Subject normalisation first: reply and forward prefixes stripped, ticket numbers removed, case folded

Exact matches on the normalised subject, plus fuzzy matches from word overlap (Jaccard) similarity over subject and body

Pairs only considered inside a configurable time window (default 7 days, boundary inclusive)

Version 1 pairs tickets from the same requester only, with a small confidence boost for repeat contacts

Outputs clusters.csv with reason and confidence per cluster, plus a repeat-contacts.md report

Fully offline and review-first: a human decides every merge

Install and usage

Requires Node.js 20 or newer. No packages to install.

git clone https://github.com/aamirtechs/ticket-dedupe.git
cd ticket-dedupe
node src/cli.js examples/tickets.csv --out-dir out

This writes clusters.csv and repeat-contacts.md to the output directory. Similarity threshold, window and repeat period are configurable.