Trust score
Optional 0–100 indicator of catalog credibility. Implementation: scripts/calculate_trust_scores.py. Background: issue #50.
Stored on YAML as trust_score and trust_score_components when calculated. Missing scores are valid — the field is optional.
Components
Owner type (5–40)
| Owner type | Score |
|---|---|
| Academy | 40 |
| Central government | 35 |
| Regional government / International | 30 |
| Local government | 25 |
| Civil society / NGO | 15 |
| Business | 10 |
| Community | 5 |
Any other type (including Federal government, Other, or missing) | 10 |
Federal government is a valid canonical owner type but is not in the score map, so it falls back to the 10-point default — as do Other and unknown/missing types.
Catalog type (−10 to +10)
| Catalog type | Score |
|---|---|
| Scientific data repository | +10 |
| Open data portal, Geoportal, Indicators catalog, Microdata catalog | +5 |
| Data marketplace | −5 |
| Data search engine (aggregator) | −10 |
| Machine learning catalog, API Catalog, Metadata catalog, Other | 0 |
| Datasets list, General research repository (unlisted) | 0 |
Types absent from the scorer map also score 0.
License / rights (−15 to +20)
The license and rights_type bonuses stack:
- Has
license_id,license_name, orlicense_url: +15 - Has
rights_typeother than null/unknown (in addition): +5 - Missing all license information: −15
rights_type: unknown: −5
Re3Data (0–20)
- Has re3data identifier: +10
- Has a trust seal (CoreTrustSeal, WDS, …): +10 more
Operational (−5 to +10)
The API and status bonuses stack:
api: trueandapi_status: active: +5status: active(in addition): +5status: inactive: −5
Formula
base = owner + catalog_type + license + re3data + operational
final = min(100, max(0, base))
| Range | Interpretation |
|---|---|
| 90–100 | Very high — certified academic / major government with licensing |
| 70–89 | High |
| 50–69 | Moderate |
| 30–49 | Low |
| 0–29 | Very low |
Recalculate
python scripts/calculate_trust_scores.py --dry-run
python scripts/calculate_trust_scores.py
Scores are heuristic. They encourage complete metadata; they are not a legal or scientific quality certificate. Full notes and examples: devdocs/trust_score_methodology.md.