Skip to main content

Preview table duplicates

Preview whole-table duplicate groups for one supported Research Table column before an approved dedupe operation.

GET/api/tables/{table_id}/dedupe/{column_key}/stats
HTTP referenceScope tables:readUpdated September 28, 2026

This page documents the canonical HTTP contract for this endpoint.

This is a read-only preview over the entire table; it does not accept row filters and does not delete rows. Empty values are preserved. The normalization strategy depends on the selected column semantics, and unsupported generic JSON identities return 422. Sample rows are capped at eight.

The returned duplicateCount and planFingerprint bind the later approved dedupe request. The registered research_table.dedupe_rows operation requires the exact preview counts and fingerprint plus approval that states the target column, rows to delete, survivor policy, and irreversibility. See Dedupe table rows for the execution endpoint and approval contract. This GET endpoint is a supporting read contract; it is not an independent delete operation.

The response example is abridged; IDs, counts, and fingerprint are illustrative.

Path parameters

  • table_idrequired

    string

    Accessible Research Table ID.

  • column_keyrequired

    string

    Key of the column whose semantic identity will be used for dedupe.

Errors

StatusMeaning
401Authentication is missing or invalid.
403The caller lacks tables:read or cannot access the table.
404The table was not found.
400The column key was not found.
422The selected column does not have a supported dedupe identity.

Docs