Exam Map & High-Yield Targets — Chapter 5
Weight: 11% — major Fundamentals domain
Must know cold
- Data modeling definition and model-as-Metadata/communication purpose.
- Four modeled data types: category, resource, business event, detail transaction.
- Entity, relationship, attribute, domain; entity vs instance.
- Cardinality vs arity; unary/binary/ternary; unary hierarchy/network.
- Foreign key; identifying/non-identifying; dependent/independent entity.
- Key taxonomy: simple, composite, compound, surrogate, super, candidate, primary, alternate, business.
- Five domain forms: type, format, list, range, rule-based.
- Six schemes: relational, dimensional, object-oriented, fact-based, time-based, NoSQL.
- Dimensional vocabulary: fact, dimension, grain, conformed dimension/fact, SCD 1/2/3, star/snowflake.
- CDM vs LDM vs PDM; canonical model; views; partitions; denormalization.
- 1NF–3NF; recognize BCNF/4NF/5NF.
- Generalization/specialization and physical subtype resolution.
- Forward vs reverse engineering and CDM/LDM/PDM build moves.
- Modeling, lineage, profiling, Metadata repository, patterns, industry models.
- Logical vs physical naming; PRISM.
- Standards, design reviews, version/change control.
- Data Model Scorecard categories and weights.
Fast distinction table
| A | B | Deciding clue |
|---|---|---|
| Data Architecture | Modeling & Design | enterprise blueprint/alignment vs precise model/design |
| Entity | Entity instance | type vs occurrence |
| Cardinality | Arity | instance participation vs number of entity types |
| Composite | Compound | 2+ attrs vs composite whose components are FKs |
| Super key | Candidate key | any unique set vs minimal unique set |
| Primary | Alternate | chosen vs unchosen candidate |
| Business | Surrogate | business meaning vs generated/no meaning |
| Identifying | Non-identifying | parent key enters child PK vs non-PK FK |
| Relational | Dimensional | operational rules vs analytics/navigation |
| Fact | Dimension | measurement vs descriptive context |
| Grain | Cardinality | meaning of fact row vs relationship participation |
| CDM | LDM | high-level scope/vocabulary vs detailed requirements |
| LDM | PDM | technology-independent vs technology-specific |
| View | Materialized view | virtual/on-demand vs stored/instantiated |
| Vertical partition | Horizontal partition | columns vs rows |
| Normalization | Denormalization | remove redundancy vs intentionally add/combine |
| Generalization | Specialization | common upward vs differences downward |
| Forward | Reverse engineering | requirements-to-physical vs database-to-conceptual |
| Pattern | Industry model | generic reusable structure vs broad industry reference |
Modeling-scheme recognition
| Scheme | Fast clue |
|---|---|
| Relational | operational business rules; normalization; one fact in one place |
| Dimensional | analytics; fact + dimensions; grain; SCD; conformance; star/snowflake |
| Object-Oriented | UML classes + attributes + operations/methods |
| Fact-Based | objects/facts/roles; controlled verbalization; no attribute construct |
| Time-Based | history; Data Vault or Anchor vocabulary |
| NoSQL | document, key-value, column, graph storage shape |
Process sequences to memorize
Forward: Requirements → CDM → LDM → PDM
Reverse: Existing database → PDM → LDM → CDM
Conceptual build: scheme → notation → initial user-view CDM → reconcile enterprise terminology → sign-off
Logical build: requirements → existing artifacts → associative entities → atomic attributes → domains → keys/relationships
Physical build: resolve abstractions → technical details/reference data → surrogate key if justified → denormalization → indexes → partitions → views/materialized structures
Lifecycle: plan → build → review → maintain
Fifteen 60-second readiness checks
- Define Data Modeling without saying “database.”
- Explain entity vs instance and cardinality vs arity.
- Classify simple/composite/compound/surrogate/super/candidate/primary/alternate/business keys.
- Name five domain forms.
- Name six modeling schemes and four NoSQL categories.
- Explain fact vs dimension and define grain.
- Recite SCD Type 1/2/3 using ORC.
- Explain CDM vs LDM vs PDM.
- Recite 1NF/2NF/3NF.
- Explain normalization vs denormalization.
- Recite forward and reverse engineering directions.
- Name the major CDM/LDM/PDM build moves.
- Recite PRISM.
- Explain standards vs design review vs version/change control.
- Reconstruct all ten Scorecard categories and the 100-point emphasis.
Answer skeleton
- Discover/analyze/scope requirements, then represent/communicate them precisely.
- Type vs occurrence; zero/one/many participation vs number of entity types.
- Construction terms describe what the key is made of; functional terms describe its identification role.
- Data type, format, list, range, rule-based.
- Relational, dimensional, OO, fact-based, time-based, NoSQL; document/key-value/column/graph.
- Fact = measurement; dimension = descriptive context; grain = one fact row’s meaning.
- Type 1 overwrite; Type 2 row; Type 3 column.
- High-level business; detailed technology-independent; technology-specific.
- Atomic/no repeats; full-key dependency; key-only dependency.
- Stabilize logic vs deliberate physical redundancy for justified need.
- Requirements→CDM→LDM→PDM; DB→PDM→LDM→CDM.
- Conceptual scope/terminology; logical attributes/domains/keys; physical technology/tuning decisions.
- Performance/ease, Reusability, Integrity, Security, Maintainability.
- Rulebook vs checkpoint vs change history.
- Requirements 15, Completeness 15, Scheme 10, Structural 15, Generic 10, Naming 5, Readability 5, Definitions 10, Enterprise consistency 5, Metadata-data agreement 10.
Question-bank weighting note
The current bank contains 72 items. Schemes are the largest cluster (14), followed by Introduction, Keys, and Relationships (7 each), then Activities and Components (6 each). The answer positions are intentionally balanced 18 A / 18 B / 18 C / 18 D.