Master Data Management: duplicate review, golden records, quality and relationships

#Purpose and scope

This chapter covers the Master Data Management module: duplicate detection and review for products, suppliers and employees, golden-record aliasing, data-quality scoring and the enterprise relationship graph.

#Where Master Data Management is in the menu

Master Data Management has no sidebar entry of its own. The module declares no sidebar group and is deliberately not a standalone application tile; /master-data-management simply redirects to its Duplicates workspace. In the source, the same three pages are declared as admin-only items of the Administration menu (Duplicate Review, Data Quality and Relationships under /administration/master-data/…), and those addresses show the same pages. If you cannot see them in the sidebar, open Duplicates, Quality or Relationships by address, and ask an administrator to check your access.

#Before you start

1. The module is admin-only. Reading candidates, quality scores and resolving an alias is open to any authenticated user of this admin module; every change needs a named permission (granted to MASTER-DATA-ADMIN): master_data_management.duplicate.scan, .duplicate.review, .duplicate.merge, .duplicate.unmerge and .relationship.manage.
2. Two separately activatable capabilities: Master Data Quality (scans, candidate queues, quality scores, history) and Master Data Governance (reviewing and merging candidates and maintaining relationships). Governance is meant for tenants that have a data-steward process.
3. The module depends on Product Management and Procurement.

#1. Scan for duplicates

1. Open Duplicates and choose Refresh assessment for the entity type: products, suppliers or employees (needs master_data_management.duplicate.scan). Other types are refused.
2. BSuit groups records by shared keys (exact identifiers such as barcode, SKU, licence, VAT, e-mail or phone, plus name-based keys) and compares only records that share a group. Pairs scoring at least 0.65 become candidates. Scores weigh exact identifiers and name similarity; employees weigh e-mail, phone and date of birth.
3. The scan also calculates a quality score and issues for each record.

#2. Review a candidate

1. Open a candidate to see the side-by-side comparison of the fields the matcher compared and the reasons it matched, such as same email or a very similar legal name.
2. Review (needs master_data_management.duplicate.review): choose Not duplicate, Confirm (confirmed duplicate) or Keep A / Keep B (merge planned, with the golden record). The golden record must be one of the two records. A suggested golden record is shown only when one side is clearly more complete; on a near tie you must choose.
3. Add notes. The history shows who decided what.

#3. Record an alias (merge) and undo it

1. On a candidate reviewed as merge planned or confirmed duplicate choose Record alias (needs master_data_management.duplicate.merge). BSuit records a reversible alias from the source record to the golden record.
2. To undo, choose Unmerge (needs master_data_management.duplicate.unmerge). The alias is deactivated, kept as an audit record, and the candidate returns to confirmed duplicate. Unmerging an already unmerged candidate does nothing.
3. Other modules can resolve a record to its golden record through the resolve lookup; it follows active aliases and is cycle-safe.

#4. Data quality

Quality lists completeness and validation issues ranked by lowest score first, for each entity type. Run an assessment to calculate scores; a record’s history shows how its score changed.

#5. Enterprise relationships

1. Open Relationships, enter a root entity type and ID, and load its graph (1 to 3 levels). Click any node to re-root the graph on it.
2. Choose Add Relationship (needs master_data_management.relationship.manage): source entity type and ID, the relationship type, and the target entity type and ID. All fields are required.