Tier 1 — Must Know Cold
Data type classification
- Reference Data: controlled domain/classification values.
- Master Data: persistent core enterprise entities.
- Transaction Data: event/activity records.
- Metadata: information describing/managing other data.
Fast test: classify/contextualize → Reference; identify entity → Master; record event → Transaction; describe source/meaning/steward/lineage → Metadata.
Authority concepts
- System of Record: authoritative creation/maintenance point.
- System of Reference: authoritative consumer access point.
- Trusted Source: governed best-available source/view.
- Golden Record: reconciled record for one entity instance; not guaranteed perfect.
Matching errors and methods
- False positive: different entities incorrectly linked/merged.
- False negative: same entity left unlinked.
- Deterministic: explicit fixed rules/patterns, repeatable.
- Probabilistic: statistical likelihood, training/tuning/thresholds.
Reconciliation workflows
- Duplicate identification: flag candidate for review; no identity/content change.
- Match-link: establish identity/cross-reference; preserve source content.
- Match-merge: reconcile attributes into unified record; requires survivorship and stronger reversal history.
Identifier management
- Global ID: enterprise identifier for reconciled entity.
- Source ID: local application identifier.
- X-Ref: mapping/history between Source IDs and Global ID.
Architecture
- Registry: enterprise index points to Master Data in source SORs.
- Transaction Hub: hub owns Master Data updates and becomes SOR.
- Consolidated: source SORs keep writes; replicated/reconciled hub becomes System of Reference.
Memory hook: Registry points; Transaction Hub owns; Consolidated copies.
Five MDM processing steps
Model → Acquire → Validate/Standardize/Enrich → Resolve Identity + Manage IDs → Share + Steward.
MDM/RDM activity skeleton
Requirements → Sources → Architecture → Model → Steward/Maintain → Governance.
Reference change flow
Request → stakeholders → impact → decision/approval → update → communicate.
Governance
Shared data needs policy and decision rights for source authority, definitions, Data Quality, matching, identifiers, changes, sharing, monitoring, stewardship, privacy/security/retention, and enterprise adoption.
Source anchor: Chapter 10, pp. 329–359.