Trust Function (TF)
A Trust Function is a serverless function that consumes the provenance tree of a resource and translates the accumulated metadata into a quantifiable trust level for the user or system.
Purpose
Trust Functions provide:
- Trust Scoring: Numerical or categorical trust ratings
- Transparency: Explanation of trust calculation
- Decision Support: Help users assess content reliability
- Automation: Enable trust-based workflows
Inputs
Trust Functions analyze:
- Provenance chains
- Agent reputation scores
- Signature verification results
- Source authority levels
- Time decay factors
- Blockchain verification status
Outputs
Trust Functions produce:
- Trust Score: Numeric (0-100) or categorical (High/Medium/Low)
- Confidence Level: Certainty of the score
- Explanation: Human-readable reasoning
- Factors: Contributing elements (positive/negative)
- Recommendations: Suggested actions
Example Trust Calculation
function calculateTrust(resource) {
let score = 0;
// Source authority (0-40 points)
if (resource.source === "EMA") score += 40;
else if (resource.source === "FDA") score += 40;
else if (resource.source === "Certified Manufacturer") score += 30;
// Provenance completeness (0-30 points)
score += resource.provenanceChain.length * 5;
// Signature verification (0-30 points)
if (resource.signatureValid) score += 30;
return {
score: Math.min(score, 100),
level: score > 80 ? "High" : score > 50 ? "Medium" : "Low"
};
}
Use Cases
Content Verification
Assess ePI or Supporting Material trustworthiness before presentation.
Lens Selection
Choose Lenses based on trust scores of their authors.
Connector Prioritization
Prefer Connectors to more trusted sources.
Alert Generation
Warn users when content has low trust scores.
Architecture
- Serverless: Function-as-a-Service deployment
- On-demand: Called when trust assessment needed
- Stateless: No persistent state between calls
- Extensible: Custom functions for specific use cases
Development
Trust Functions can be:
- Developed by FOSPS community
- Customized per deployment
- Validated against test datasets
- Inspected via CTF Inspector
Related Concepts
- CTF - Trust framework
- Provenance - Input data
- CTF Inspector - Development tool
- ePI - Scored content
- Lens - Scored algorithms
- FOSPS - Platform architecture