
Walk through the four-stage curation lifecycle in Ataccama ONE - load, deduplicate, remediate, and publish - and learn how this governed process turns messy source records into a trusted golden record.
Create a ONE Data table, map source columns to schema attributes, set a primary key and display name, and load your first records ready for deduplication.
Configure exact, fuzzy, and phonetic match keys, run the deduplication engine, and review match proposals with confidence scores to confirm or reject candidate duplicates.
Follow a record from a failing DQ rule into the remediation queue, apply and submit a manual correction, and trace it through Pending Approval into the audit log.
Open the compare view for an accepted match proposal, select the winning value for each conflicting field, and execute the merge that produces your golden record.
Run post-remediation validation against your golden records, read the validation report, and distinguish failures needing rework from those needing rule adjustment.
Publish golden records through the ONE Data API, a Desktop plan export, or a virtual catalog item, then confirm delivery and monitor downstream data quality.
This course contains the use of artificial intelligence.
Monitoring data quality identifies problems. Curation fixes them. If your organisation has datasets with duplicate, incomplete, or incorrect records that need to be resolved before data can be trusted, this course gives you the practical skills to do it in Ataccama ONE.
You will learn the full curation lifecycle from raw data through to a validated, published golden record. Starting with loading data into ONE Data Tables, you will configure deduplication to find and match duplicate records, use remediation workflows to review and correct individual records that fail quality checks, compare candidates and merge them into a single authoritative golden record, validate the results, and publish clean data back to your source systems or downstream consumers.
What you will build by the end of this course:
- A configured ONE Data Table with a loaded dataset ready for curation
- A deduplication configuration that identifies matching and duplicate records
- A remediation workflow for human-in-the-loop record correction
- A compare-and-merge step that produces a golden record for each entity group
- Validation checks on the golden record set
- A published output of clean, curated data for downstream use
Prerequisites: UDY03 and UDY06.
What to take next: UDY07 (Stop Bad Data at the Source) to prevent the problems that make curation necessary.