avra-cadavra/.cursor/skills/expertise-system-implementation

Expertise System Implementation

Core Principle

Users can become experts through multiple paths. Expertise is recognized at different geographic levels.

World Model Transition Notice

IMPORTANT: The hardcoded weights below (40%, 25%, 20%, etc.) are scheduled for replacement by the energy function per Master Plan Phase 4.3 (Expert System Integration). The multi-path structure and geographic hierarchy are preserved as input features for the world model's state/action encoders.

When implementing expert features:

• ✅ Wire expert reputation data (golden expert score, expertise level, path scores) as state encoder features
• ✅ Wire geographic hierarchy data (neighborhood, locality, city) as action encoder features
• ✅ Capture expert event outcomes for episodic memory (Phase 4.3.3)
• ❌ Do NOT create new hardcoded weight combinations for expert scoring
• ❌ Do NOT hardcode expert priority calculations

Expert Discovery: ExpertiseMatchingService._getAllUsers() currently returns empty -- Phase 8.4 adds mesh-based expert discovery via BLE advertisements.

Reference: docs/MASTER_PLAN.md Phase 4.3, Phase 8.4

Multi-Path Expertise

Six Paths to Expertise

1. Exploration (40%) - Visits, reviews, check-ins, dwell time
2. Credentials (25%) - Degrees, certifications, published work
3. Influence (20%) - Followers, shares, list curation (logarithmic normalization)
4. Professional (25%) - Proof of work, roles, peer endorsements
5. Community (15%) - Questions answered, events hosted, contributions
6. Local (varies) - Locality-based expertise with golden expert bonus

Weighted Calculation

/// Calculate weighted expertise score
double calculateExpertiseScore({
  required double explorationScore,
  required double credentialsScore,
  required double influenceScore,
  required double professionalScore,
  required double communityScore,
  required double localScore,
}) {
  return (explorationScore * 0.40) +
         (credentialsScore * 0.25) +
         (influenceScore * 0.20) +
         (professionalScore * 0.25) +
         (communityScore * 0.15) +
         (localScore * localWeight);
}

Geographic Hierarchy

Hierarchy Levels

Local → City → State → National → Global → Universal

Local Expertise

/// Calculate local expertise
Future<LocalExpertise> calculateLocalExpertise({
  required String userId,
  required String locality,
}) async {
  // Calculate expertise at locality level
  final visits = await _getVisitsInLocality(userId, locality);
  final contributions = await _getContributionsInLocality(userId, locality);
  
  return LocalExpertise(
    locality: locality,
    score: _calculateLocalScore(visits, contributions),
    goldenExpert: await _checkGoldenExpert(userId, locality),
  );
}

Dynamic Thresholds

/// Get effective requirements (adjusted for phase + saturation)
ThresholdValues getEffectiveRequirements({
  required ExpertiseRequirements requirements,
  required PlatformPhase platformPhase,
  required SaturationMetrics saturationMetrics,
}) {
  // Get phase multiplier
  final phaseMultiplier = platformPhase.getCategoryMultiplier(requirements.category);
  
  // Get saturation multiplier
  final saturationMultiplier = saturationMetrics.getSaturationMultiplier();
  
  // Calculate total multiplier
  final totalMultiplier = phaseMultiplier * saturationMultiplier;
  
  // Apply multiplier to thresholds
  return requirements.thresholdValues.applyMultiplier(totalMultiplier);
}

Reference

• lib/core/services/expertise_calculation_service.dart
• lib/core/services/multi_path_expertise_service.dart
• docs/patents/category_3_expertise_economic_systems/01_multi_path_dynamic_expertise/