wehub-resource-sync
7.0 KiB
UK Government Data Quality Framework — ArcKit Reference Guide
Guide Origin: Official | ArcKit Version: [VERSION]
This guide maps the Government Data Quality Framework (DQF) to ArcKit commands and artefacts. The DQF is published by the Government Data Quality Hub and provides principles, dimensions, and practical tools for managing data quality across government.
Five Principles
| # | DQF Principle | Description | ArcKit Evidence |
|---|---|---|---|
| 1 | Commit to data quality | Establish accountability and ongoing assessment | /arckit:data-model (Data Quality Framework section — owners, targets, monitoring) |
| 2 | Know your users and their needs | Research quality requirements of data consumers | /arckit:stakeholders (data governance roles), /arckit:data-mesh-contract (consumer SLAs) |
| 3 | Assess quality throughout the data lifecycle | Monitor at every stage: collect, store, use, share, archive | /arckit:data-model (quality metrics per entity, lifecycle stages) |
| 4 | Communicate data quality clearly | Transparent, plain-language quality information for consumers | /arckit:data-mesh-contract (quality statements, SLA targets, consumer documentation) |
| 5 | Anticipate changes affecting quality | Plan proactively to prevent quality degradation | /arckit:risk (data quality risks), /arckit:operationalize (monitoring, alerting) |
Six Quality Dimensions
These dimensions are already scaffolded in the /arckit:data-model template with per-entity targets, validation rules, and measurement methods.
| Dimension | Definition | Template Section |
|---|---|---|
| Completeness | All required records and values are present | Quality Dimensions → Completeness |
| Uniqueness | No unnecessary duplication of records | Quality Dimensions → Uniqueness |
| Consistency | Values align across systems and don't contradict | Quality Dimensions → Consistency |
| Timeliness | Data reflects current information, available when needed | Quality Dimensions → Timeliness |
| Validity | Data conforms to expected formats, ranges, and rules | Quality Dimensions → Validity |
| Accuracy | Data correctly represents real-world conditions | Quality Dimensions → Accuracy |
Four Practical Tools
1. Data Quality Action Plans
Prioritised improvement steps for critical data issues. The data-model template captures this through:
- Quality targets per entity and dimension (gap = target vs current)
- Issue classification (Critical/High/Medium/Low)
- Resolution process with owner assignment
When to create a formal action plan: When quality scores consistently fall below targets, or when a new data source is onboarded with unknown quality characteristics.
2. Root Cause Analysis
Techniques for addressing underlying data quality problems rather than symptoms.
| Technique | When to Use |
|---|---|
| 5 Whys | Simple causal chains — "why is email accuracy dropping?" |
| Fishbone (Ishikawa) | Multiple contributing factors — people, process, technology, data sources |
| Pareto Analysis | Prioritise — which 20% of causes drive 80% of quality issues? |
ArcKit integration: Record root causes and remediation in /arckit:risk (risk register) and track actions in /arckit:backlog.
3. Metadata Guidance
Minimum metadata set for documenting data characteristics. The data-model template captures this through:
- Entity catalogue (definitions, data types, keys, constraints)
- Data dictionary with attribute-level descriptions
- Source system and refresh cadence per entity
- Data steward contact per entity/domain
4. Data Maturity Model
Self-assessment of organisational data quality capability.
| Level | Description | Indicators |
|---|---|---|
| Initial | Ad hoc, reactive quality management | No formal ownership, quality issues discovered by users |
| Repeatable | Basic processes and ownership defined | Data stewards assigned, some quality rules |
| Defined | Standardised processes across the organisation | Quality dimensions measured, dashboards in place |
| Managed | Quantitative quality management with targets | SLAs defined, automated monitoring, regular reporting |
| Optimising | Continuous improvement, predictive quality | Proactive issue prevention, root cause analysis embedded |
ArcKit evidence: The data-model template's quality metrics section (overall score, monitoring, alerting) provides evidence for Defined/Managed maturity. The issue resolution process supports Managed/Optimising.
Data Lifecycle Stages
The DQF expects quality assessment at every stage of the data lifecycle.
| Lifecycle Stage | Quality Focus | ArcKit Artefact |
|---|---|---|
| Plan | Define quality requirements and targets | /arckit:requirements (DR-xxx data requirements) |
| Collect / Ingest | Validate at point of entry | /arckit:data-model (validation rules, reject/accept logic) |
| Prepare / Store / Maintain | Cleanse, deduplicate, reconcile | /arckit:data-model (deduplication rules, reconciliation process) |
| Use / Process | Monitor quality during processing | /arckit:data-model (quality metrics, dashboards) |
| Share / Publish | Communicate quality to consumers | /arckit:data-mesh-contract (SLAs, quality statements) |
| Archive / Destroy | Maintain quality of retained data | /arckit:data-model (retention policy, disposal procedures) |
Relationship to Other Frameworks
| Framework | Relationship to DQF |
|---|---|
| National Data Strategy | DQF implements the Data Foundations pillar (Mission 3: transforming government data use) |
| GovS 010: Analysis | Parent functional standard for analytical quality and data management |
| ISO 8000 | International data quality standard — DQF dimensions align with ISO 8000 |
| DAMA DMBOK | Industry data management body of knowledge — DQF covers a subset of DAMA quality domains |
| AI Readiness Guidelines | AI-ready datasets require DQF-level quality assurance |