Duplicate Activity Detection

NHF manages 1 400 local chapters where multiple coordinators can register activities for the same contact, creating a concrete risk of double-counting. Duplicated records inflate Bufdir-reported activity metrics, potentially constituting grant misreporting. For organizations receiving government funding based on activity counts, duplicate detection is a compliance safeguard. Catching duplicates before the Bufdir export cycle saves significant manual reconciliation effort and protects organizational credibility with government funders. The feature also reduces coordinator workload by surfacing likely duplicates rather than requiring manual cross-referencing across full activity history.

Detection runs as a scheduled background job comparing newly submitted activities against recent history on dimensions including peer mentor ID, contact ID, activity type, and date within configurable tolerances (e.g. same contact + same type + same date ± 1 day = high-confidence duplicate). A similarity score (0-100) and duplicate_candidate_id are stored on the activities table. High-scoring pairs are auto-flagged via activity-flagging with reason_code duplicate_suspected. The admin UI adds a Duplicates tab showing pairs ranked by score. Administrators can confirm a duplicate (soft-delete one record) or mark false positive (clears flag and score). All decisions are audit-logged and configurable per organization.