SemanticGIS

ANALYSIS-SUMMARY

4 min read

Dataset-to-Leaf Analysis Complete

Analysis Date: April 18, 2026 Status: Analysis only - no new leaves created (per user request)

Summary of Work Completed

1. Forward Matching: Leaves → Datasets

Status: ✅ Previously completed (second pass, 153/200 realisations matched)

  • All 46 leaves updated with “Realised By Links” sections
  • Unmatched realisations visible in each leaf for manual curation
  • Markdown lint verified clean

2. Reverse Matching: Unmatched Datasets → Leaves

Status: ✅ Just completed

Results:

  • Total unmatched datasets: 2,069
  • Matched to existing leaves: 317 (15.3%)
  • Remaining unmatched: 1,752 (84.7%)

Top leaves receiving matches:

  1. Groundwater Chemistry - 78 datasets
  2. Nature Protection Areas - 27 datasets
  3. Protected Water Use Zones - 23 datasets
  4. Fisheries Intensity - 20 datasets
  5. Groundwater Bodies - 19 datasets

Output files generated:

  • datasets-matched-to-leaves.json (317 matches with confidence scores)
  • datasets-matched-to-leaves.md (human-readable match table)
  • datasets-remaining-unmatched.json (1,752 unmatched with metadata)
  • datasets-remaining-unmatched.md (analysis table)

Gap Analysis: Potential New Leaves

Status: ✅ Analysis complete - awaiting human consent before creation

Top Categories of Unmatched Datasets (by cluster):

ClusterCountPriorityStatus
INSPIRE/Open Data Services403LowInfrastructure/platform issues, not subject-specific
Other/Uncategorized847Needs reviewDiverse datasets, may group into multiple leaves
Environmental Noise98HIGHEU Noise Directive requirement; 98 datasets ready to map
Land Planning (PlanDK)65HIGHCritical INSPIRE theme; may need 2-3 separate leaves
Water Source Protection53HIGHDrinking water infrastructure; distinct from existing water leaves
Fisheries Data58Medium-HighPartly covered; additional datasets available
Forest Management38Medium-HighDistinct from habitat-extent; management-focused
Agricultural Practices29MediumCould extend existing Agriculture leaf
Biodiversity Directives (Natura 2000)48Medium-HighComplements existing Nature Protection Areas leaf
Transport Infrastructure21MediumRail, road, pipeline networks
Emissions & Air Quality18MediumRegulatory; complements existing Atmospheric Deposition leaf
Climate-Related Land Use12MediumPeatland management, forest restoration
Cultural Heritage & Archaeology13Low-MediumOptional; not core geospatial data
Greenland Datasets17LowScope question: include Greenland data?
Other Clusters<15 eachLowIndustrial facilities, waste, shipping, research monitoring

Tier 1: Immediate Value (High priority)

  1. Environmental Noise & Air Transport (98 datasets)

    • EU Noise Directive 2002/49/EC compliance
    • Noise mapping by source: rail, road, aviation
    • Sample datasets: “Støj fra jernbaner”, “Støj fra større vejanlæg”

  2. Spatial Planning & Municipal Decisions (65 datasets)

    • PlanDK registry datasets
    • Sample datasets: “Store husdyrbrug - PlanDK”, “Administrative skel - PlanDK”
    • Consider: May warrant separate leaves for municipal/regional/national levels

  3. Drinking Water & Water Source Protection (53 datasets)

    • Distinct from existing groundwater/freshwater leaves (focus on infrastructure)
    • Sample datasets: “VP3 - Dybe drikkevandsforekomster”, “Sensitive Drinking Water Areas”

Tier 2: Moderate Value (Medium priority)

  1. Natura 2000 & EU Biodiversity Directives (48 datasets)

    • Complements “Nature Protection Areas” leaf
    • Sample: “Natura 2000 Bird Protection Areas”, EU habitat directive datasets

  2. Forest Ecosystems & Management (38 datasets)

    • Distinct from “Habitat Extent” (focus on management/forestry)
    • Sample: “Forest Type Map”, “Management units”, “Urørt skov”

  3. Fisheries Monitoring & Management (58 datasets)

    • Partly overlaps existing “Fisheries Intensity” leaf
    • Could extend existing leaf with additional VP/NP datasets

  4. Agricultural & Land Use Practices (29 datasets)

    • Ecological condition, crop rotation, soil management
    • Could extend existing “Agricultural Land Management” leaf

Tier 3: Optional/Low Priority

  1. Air Quality & Emissions Inventories (18 datasets)

    • Regulatory reporting (PM10, NOx); could extend “Atmospheric Deposition” leaf

  2. Industrial Facilities & Contamination Risk (10 datasets)

    • Could extend “Spill and Pollution Incidents” leaf

  3. Transport Infrastructure Networks (21 datasets)

    • Rail corridors, pipelines, utility lines

What Requires Human Review

1. INSPIRE/Service Issue (403 datasets)

Many unmatched entries are metadata about INSPIRE web services rather than actual datasets:

  • “DK INSPIRE AM Areas for Groundwater Protection - service”
  • “DK INSPIRE PS Natura 2000 - service”

Action needed: Determine if these should be:

  • Excluded from analysis (metadata artifacts)?
  • Merged with parent dataset entries?
  • Handled differently?

2. Large “Other” Category (847 datasets)

Nearly 50% of unmatched datasets remain uncategorized. Likely causes:

  • Diverse theme combinations not captured by keyword clustering
  • Multi-language variations (Danish/English)
  • Technical/platform datasets (versions, updates)

Action needed: Manual sampling to identify patterns

3. Greenland Datasets (17)

  • Question: Should SemanticGIS cover Greenlandic data?
  • If yes, may need separate organizational structure or scope clarification

Files Available for Review

All analysis outputs in: /DanishData/content/assets/lookup/

  • datasets-matched-to-leaves.md – Human-readable matches (317 datasets mapped to 46 leaves)
  • datasets-remaining-unmatched.md – Top 100 unmatched datasets with best-match suggestions
  • gap-analysis-potential-new-leaves.md – Detailed cluster analysis with recommendations

Next Steps (Awaiting Your Guidance)

Option A: Review & Create New Leaves

  1. Review gap analysis and prioritized recommendations
  2. For each approved new leaf:
    • Define: concept, question, primary_collection
    • Map realisations from unmatched datasets list
    • Follow existing leaf template in DanishData/content/Leaves/
  3. Notify when new leaves created; re-run matching script
  4. Update leaves with “Realised By Links” sections

Option B: Refine Matching Heuristics

  1. Tune F1 threshold or keyword matching logic
  2. Handle INSPIRE service metadata separately
  3. Improve language normalization for Danish/English

Option C: Manual Curation

  1. Review datasets-remaining-unmatched.json manually
  2. Create mapping spreadsheet of dataset → recommended leaf
  3. Batch create new leaves based on manual analysis

No action until you confirm which path forward.

Bidirectional Coverage Summary

DirectionStatusDetails
Leaves → Datasets✅ 153/200Realisations matched to concrete datasets
Datasets → Leaves✅ 317/2069Unmatched datasets analyzed; gaps identified
Visibility✅ CompleteLeaves have “Realised By Links” sections; unmatched lists in audit artifacts
New Leaves⏸️ PendingGap analysis complete; awaiting human consent

Graph View