Publish a dataset
Take a lakehouse table through the intake pipeline so other teams can find it, request it and build on it.
For: Data Steward · 1 min read
Prepare
- The table is in the lakehouse and registered in Dataplex, with an owner and a refresh schedule.
- Columns are described. Customer identifiers are tokenised; MyKad numbers never appear in clear.
- You know its classification and which sensitivity tags apply.
Steps
- 1Open Publish and choose Dataset.
- 2Fill the metadata: name, business domain, classification, reuse terms and usage rights.
- 3Complete the responsible AI checklist. Answers are attached to the dataset card and audited.
- 4Submit. The pipeline runs a metadata check, a security and PII scan, and the checklist review.
- 5A reviewer signs off. If the PII scan flags anything, you as data owner must sign off too.
- 6The dataset goes live in the catalogue with a Data Health Score.
After publishing
Keep the refresh cadence you declared: the Data Steward Agent flags late refreshes and quality drops to you and to the consuming teams.
Read next
Was this helpful?