Battle Rounds — Complete Answer & Rationale Key
Round 1 — Rapid Classification
- Data Governance. Deciding clue: authoritative definition/conflict decision. Data Management would implement the approved result. Sources
- Data Management. Deciding clue: configure/operate approved definition and mapping. Sources
- Data Owner. Deciding clue: final business-domain accountability. Source
- Business Steward / Data Steward in the business-SME sense. Deciding clue: terms, valid values and business rules. Sources
- Data Quality Analyst. Deciding clue: condition/trend/root-cause analysis. Source
- Data Architecture. Deciding clue: enterprise target/alignment/reuse. Sources
- Data Modeling & Design. Deciding clue: cardinality/keys/normalization. Sources
- Metamodel. Deciding clue: entities are Metadata objects. Source
- Business Glossary. Deciding clue: approved business meaning. Source
- Data Catalog/Directory. Deciding clue: discovery/location. Source
Round 2 — Changed-Fact Switches
- implements / operates. Governance → Management because authority becomes execution.
- defines / maintains. Owner → Steward because final accountability becomes business semantics/rules work.
- monitors/analyzes. Steward → DQ Analyst because expectations become condition/root-cause analysis.
- cardinality/relationships. Architecture → Modeling because enterprise blueprint becomes detailed representation/design.
- Metadata. Data model → metamodel because the modeled entities now describe data/systems/management context.
- structural/technical properties (for example datatype/nullability/uniqueness). Glossary → Dictionary because meaning becomes field structure.
- Customer / the business term (or another term). Catalog → Glossary because discovery becomes business meaning.
- Customer/entity. Reference → Master because controlled-code translation becomes persistent entity identity.
- ongoing/recurring. Profiling → Monitoring because discovery becomes repeated control over time.
- process/cause/mechanism. Correction → Root-Cause Remediation because existing-data repair becomes elimination of recurrence.
Source set: the linked Confusion Book entries for each pair.
Round 3 — Best-Answer Elimination
21 — B. Accuracy
The value conforms to format/domain rules, so validity is not the failure. It represents the wrong real-world customer, which makes accuracy the best answer. Completeness asks presence; uniqueness asks one representation per real entity.
Source: Validity vs Accuracy.
22 — C. Validity
The explicit clue is violation of an approved range. Accuracy could also be false in reality, but the stem gives a rule-conformance failure, making validity the best answer. Integrity is relationship coherence; consistency is expected agreement.
Source: Validity vs Accuracy.
23 — B. Reference Data Management
The problem is controlled country-code values and their mapping/crosswalk. MDM resolves entity identity; Modeling describes structures; Monitoring watches DQ over time.
Source: Reference vs Master.
24 — C. Master Data Management
The decisive question is whether records across sources represent the same persistent Customer entity. Reference Data controls classifications/codes; Glossary manages meaning; Architecture provides enterprise blueprint/alignment.
Source: Reference vs Master.
25 — A. Data profiling
Null frequencies/value distributions are profiling techniques used to reveal patterns. Monitoring is recurring; RCA/correction are downstream improvement work.
Source: Profiling vs Assessment vs Monitoring.
26 — B. Initial Data Quality assessment
The profiling findings are now being judged as a baseline and prioritized, which is the management activity of initial assessment. Profiling supplied evidence; monitoring is ongoing.
Source: Profiling vs Assessment vs Monitoring.
27 — C. Monitoring
Recurring hourly trending plus threshold alerts is ongoing monitoring. Initial assessment establishes baseline; correction repairs data; reference control is unrelated to the recurrence clue.
Source: Profiling vs Assessment vs Monitoring.
28 — B. Prevention
Input validation stops future bad values at creation. Correction fixes existing records; monitoring observes; RCA investigates/removes underlying causes but the described control itself is preventive.
Source: Prevention vs Correction vs RCA.
29 — C. Correction
The source defect is already fixed and the task is repairing existing invalid data. That is correction. Prevention stops new defects; monitoring observes; harvesting is unrelated.
Source: Prevention vs Correction vs RCA.
30 — B. Root-cause remediation
The team changes the underlying workflow/responsibility that repeatedly creates the defect. Correction would only repair existing addresses; profiling/validity measurement diagnose but do not remove the cause.
Source: Prevention/Correction/RCA.
Repair rule
If an answer was wrong or slow, return to the linked Confusion Book entry, restate the deciding clue, then create one changed-fact version that makes your wrong option correct. Only repeated real misses belong in later Artifact 08 work.