Data quality and context coverage
Quality rule, measurement, violation, coverage and remediation plan. Source completeness, field fill rate and truth of facts are different indicators.
Bundle → Layer → Finding → Questions Filled
3 bundles · 3 layers · 5 findings · 10 questions
Rules and measurement What quality means and how it was measured.
Quality rules
Rules for distinct dimensions of quality.
Rule definition
A rule for one dimension such as completeness, accuracy or freshness.
- Which quality dimension does the rule measure?
- For which purpose is the freshness or accuracy threshold set?
Evaluation
A measurement that pins the rule version and the sample.
- Which rule version and sample did the evaluation use?
- Does a field fill rate get reported as accuracy?
Coverage How much of the relevant world the data covers.
Coverage assessment
Source coverage with an explicit denominator or its absence.
Denominator
Whether the full population is known.
- Is the denominator for coverage known, estimated or unknown?
- Is an unknown denominator stated explicitly in the result?
Issues and remediation Violations found and how they are fixed.
Quality issues
Violations with their owner and remediation plan.
Issue record
A violation with impact, owner and plan.
- Which records and consumers does the violation affect?
- Who owns the remediation, and by when?
Assessment history
Corrections that do not rewrite past assessments.
- Was a past assessment kept unchanged after the data was corrected?
- Does a new assessment exist for the corrected data?
Classifiers Filled
- Family
- Enterprise profiles
- Category
- Enterprise subject
- Entry kind
- subject
- Domain
- EnterpriseDAT
- Industry
- Cross-industry
- Tags
- EM-DAT-04W2subjectDataQualityRuleQualityEvaluationQualityIssueCoverageAssessment
What it is Filled
Quality rule, measurement, violation, coverage and remediation plan. Source completeness, field fill rate and truth of facts are different indicators.
Why it exists Filled
Quality rule, measurement, violation, coverage and remediation plan. Source completeness, field fill rate and truth of facts are different indicators.
Distinguishing features Filled
- Completeness, field fill rate, accuracy and freshness are separate indicators, never one score.
- An assessment pins the rule version and the sample it was made on.
- When the population size is unknown, coverage states this rather than guessing a percentage.
- A correction produces a new assessment and leaves earlier ones as they were.
What robots and AI may and may not do Filled
Must not
- Declare a 100 percent field fill rate to be 100 percent accuracy.
- Report coverage as a percentage when the denominator is unknown.
- Rewrite a past assessment after the data has been corrected.
- Suppress failing quality results from reports to consumers.
Only with a human decision
- Accepting known quality defects for data used in decisions about people.
- Changing quality thresholds for a data contract.
May
- Run quality rules and record results with rule version and sample.
- Report each quality dimension separately for a dataset.
- Open issues for violations with affected records and consumers.
Moral aspects Filled
- Poor data quality can lead to unfair or harmful decisions about individuals.
- Overstated quality misleads consumers and shifts risk onto them.
Who is affected
- Data consumers
- People described by the data
- Data owners
Owners Filled
Steward
The data owner or head of analytics who answers for data assets and their fitness for use.
Master systems
- Data catalogue
- Data warehouse
- BI platform
- Data orchestrator
Links to other meta-models Filled
neighbor
- EM-DAT-01
references
- WM-DAT-007 - conceptual-candidate
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- Identified by rule identifier and version, and each evaluation by dataset version, rule version and time.
Direct properties not applicable Not applicable
Not applicable
Enterprise record contour: physical properties belong to referenced world models.
Recognition optional Filled
- A quality record names a dimension, a rule, a sample and a result.
- Often confused with a single overall quality score or a fill-rate report.
Capabilities and actions required Filled
- Quality can be compared across dataset versions on the same rule.
- Fitness for a specific purpose can be judged dimension by dimension.
Hazards and failure modes required Filled
- Misleading single scores that mix dimensions.
- Invented coverage percentages without a known population.
- Lost history when assessments are overwritten.
Standards and interfaces required Filled
- ISO/IEC 25012 data quality model.
- ISO 8000 series on data quality.
- W3C Data Quality Vocabulary (DQV).
Context of use required Filled
- Used wherever data feeds decisions, reports or AI models.
- Regulated reporting and AI training data have explicit quality expectations.
Sources Filled
- W3C DCAT: dataset, distribution, catalog
- W3C PROV and Data Cube: provenance and observation
- SDMX and BI/data catalog practice: indicators, breakdowns, definitions and releases
- W3C Data Quality Vocabulary (DQV) - Checked live (HTTP 200) on 2026-10-07.
- W3C PROV-O: The PROV Ontology - Checked live (HTTP 200) on 2026-10-07.
- W3C The RDF Data Cube Vocabulary - Checked live (HTTP 200) on 2026-10-07.
Open questions
- How does completeness differ from accuracy?
- How to measure coverage without a known denominator?
- What does freshness mean for a specific purpose?
- Установить границу и решение reuse/extend/new по действующим спецификациям.
- Подтвердить semantic crosswalk, права и source mastership.
- Выбрать immutable refs; провести проверки fixtures до заявления о публикационной готовности.
Machine files
Provenance
enterprise research programme · published-partial
Built from: enterprise/models/em-dat-04/brief.json, ver-cy/world-models/card-supplements/em-dat-04.json
Still in Russian: suggested owner, blocking decisions, vercy candidates. Translate in research/enterprise/i18n/units.en.json.