# Northstar MDM Workshop — Part 1

This package supports the Data4AI tutorial “Why AI Makes Master Data Management Non-Negotiable.” Use it only in an approved non-production EBX 6.x environment.

## Import order

1. Import `northstar_master.xsd` as the embedded semantic model `northstar_master`.
2. Publish a reviewed workshop model snapshot.
3. Create dataset `northstar_lab` in child dataspace `AI_MDM_WORKSHOP`.
4. Import `SourceSystem.csv`.
5. Import `Customer.csv`.
6. Import `Product.csv`.

Run EBX simulation or preview before every data import. The records are intentionally inconsistent and must not be normalized during ingestion.

## What the seed data demonstrates

- Profiling reveals missing, inconsistent and differently formatted source values.
- EBX model constraints and EBX Script evaluate declared validation requirements.
- The separately installed Match and Merge add-on generates and resolves possible duplicate groups.
- A role-specific perspective can bring validation results, profiling/DQ findings and matching stewardship together without copying them into a generic observation table.

Do not treat a normalized comparison value as a mastered identity. Matching, survivorship and steward decisions are separate governed operations.

All records are synthetic. The `.test` email domain is reserved for examples and must not be replaced with personal information.