DFS Pro Review Queue

The Review Queue is the human review layer for DFS Pro. Use it to resolve uncertain fusion results and rejected ingest rows before downstream workflows consume the data.

Open:

Data Integration > Review Queue

The page has two working areas:

• review items;
• rejected rows.

Review items

Review items are created when a workflow needs a human decision.

Common reasons:

Reason	Meaning
Conflict	Two or more sources disagree or map to competing outputs.
Source disagreement	Source systems provide inconsistent values or identities.
Low confidence	Matching or fusion confidence is below the review threshold.
Manual flag	A user or process marked the result for review.

Common statuses:

Status	Meaning
Pending	Waiting for reviewer action.
Approved	Reviewer accepted the item.
Rejected	Reviewer rejected the item.
Manually resolved	Reviewer fixed or resolved it outside the standard approve/reject path.

Resolve a review item

1. Open Review Queue.
2. Filter to Pending.
3. Filter by reason if the queue is large.
4. Open a review item.
5. Read summary and context.
6. Compare source records, timestamps, keys, and confidence.
7. Add a note.
8. Choose Approve, Reject, or Manually resolve.
9. Check the action log.

Use comments when the item needs more context before a final decision.

Rejected rows

Rejected rows come from ingest validation. They represent source rows or records that DFS could not accept.

Common states:

State	Meaning
Pending	A row is waiting for triage.
Acknowledged	A reviewer has seen the row.
Fixed in source	The upstream source was corrected.
Reprocessed	A source-reread reprocess was requested and completed.

Rejected rows can include:

• source;
• dataset ID;
• connector ID;
• reason;
• row index;
• source file;
• raw payload;
• correlation ID.

Resolve rejected rows

1. Open the rejected rows tab.
2. Filter to Pending.
3. Open a row.
4. Inspect reason, source, dataset, connector, row index, and raw payload.
5. Acknowledge the row.
6. Fix the upstream source if needed.
7. Mark the row Fixed in source.
8. Reprocess with source-reread when connector provenance is available.
9. Check replay evidence and sync result.

Source-reread reprocess is useful when the upstream data has been corrected and the row can be ingested again from the original connector.

Review notes

Good review notes include:

• what was checked;
• which source was trusted;
• why the item was approved, rejected, or manually resolved;
• whether upstream data was corrected;
• follow-up owner when needed.

Before downstream use

Before a fusion output or dataset is used downstream, confirm:

• pending review items are resolved or accepted as known limitations;
• rejected rows are reviewed for critical sources;
• replay evidence is recorded when rows are reprocessed;
• output dataset preview and profile are acceptable;
• data steward is assigned when the data will be reused.

Related pages

Page	Use
Fusion Tasks	Understand how review items are created by fusion workflows.
DFS Pro Datasets	Validate output datasets after review.
Data Quality	Check connector health before reviewing rejected rows.