Skip to content

digital-twin.md

Latest commit

 

History

History
134 lines (103 loc) · 4.53 KB

digital-twin.md

File metadata and controls

134 lines (103 loc) · 4.53 KB

Digital Twin Personality Enhancement

Quantitative personality modeling and prediction system that accurately embodies a human's values, decision patterns, and communication style.

Vision

Transform the Digital Twin from a document capture system into a quantitative personality modeling and prediction system.

Architecture

  • Digital Twin Service (server/services/digitalTwin.js): Trait analysis, confidence scoring, gap recommendations
  • Digital Twin Routes (server/routes/digital-twin.js): REST API endpoints
  • Digital Twin Validation (server/lib/digitalTwinValidation.js): Zod schemas for trait data

Features

Phase 1: Quantitative Personality Modeling (Complete)

Big Five Trait Scoring

  • Quantified OCEAN scores (Openness, Conscientiousness, Extraversion, Agreeableness, Neuroticism)
  • Infer scores from existing documents using LLM analysis
  • Allow manual override/adjustment
  • Store in meta.json under traits.bigFive

Values Hierarchy

  • Extract explicit values from VALUES.md and NON_NEGOTIABLES.md
  • Create ranked values list with conflict resolution rules
  • Store in meta.json under traits.valuesHierarchy

Communication Fingerprint

  • Quantify writing style: formality (1-10), verbosity (1-10), emoji usage, sentence length avg
  • Extract from WRITING_STYLE.md and writing samples
  • Store in meta.json under traits.communicationProfile

Phase 2: Personality Confidence Scoring (Complete)

Coverage Metrics

  • For each Big Five dimension: evidence count from documents
  • For each value: supporting document count + specificity score
  • For communication: sample diversity, consistency across samples

Confidence Algorithm

confidence(aspect) = min(1.0,
  (evidence_count / required_evidence) *
  (consistency_score) *
  (recency_weight)
)

Gap Recommendations

  • Identify lowest-confidence aspects
  • Generate specific questions to fill gaps
  • Prioritize enrichment categories by confidence gap

Phase 4: External Data Integration (Complete)

Import from external sources to reduce manual input:

  • Goodreads CSV import for reading preferences
  • Spotify/Last.fm for music profile
  • Calendar pattern analysis for routines

Data Structure

traits: {
  bigFive: { O: 0.75, C: 0.82, E: 0.45, A: 0.68, N: 0.32 },
  valuesHierarchy: ["authenticity", "growth", "family", ...],
  communicationProfile: {
    formality: 6,
    verbosity: 4,
    avgSentenceLength: 18,
    emojiUsage: "rare",
    preferredTone: "direct-but-warm"
  },
  lastAnalyzed: "2026-01-21T..."
}

UI Components

  • PersonalityMap.jsx - Radar chart of Big Five with confidence coloring
  • ConfidenceGauge.jsx - Per-dimension confidence indicator
  • GapRecommendations.jsx - Prioritized enrichment suggestions
  • TraitEditor.jsx - Manual trait override interface

API Endpoints

Route Description
GET /api/digital-twin/traits Get all trait scores
POST /api/digital-twin/traits/analyze Analyze documents to extract traits
PUT /api/digital-twin/traits/:category Manual override trait scores
GET /api/digital-twin/confidence Get confidence scores
POST /api/digital-twin/confidence/calculate Recalculate confidence
GET /api/digital-twin/gaps Get gap recommendations

Planned Phases

Phase 3: Behavioral Feedback Loop

  • Response validation: "sounds like me" / "not quite me" ratings
  • Feedback analysis and document improvement suggestions
  • Adaptive document weighting based on feedback patterns

Phase 5: Multi-Modal Personality Capture

  • Voice analysis for speech patterns
  • Video interview for facial expressions and gestures
  • Comparison of spoken vs written style

Phase 6: Advanced Behavioral Testing

  • Complex multi-turn conversation scenarios
  • Ethical dilemma tests aligned with stated values
  • Quantitative scoring of communication style match
  • Adversarial testing of boundaries

Phase 7: Twin Personas & Context Switching

  • Named personas (Professional, Casual, Family, Creative)
  • Blending rules for trait variation per context
  • Per-persona testing

Success Metrics

Metric Current Target
Behavioral test pass rate ~70% >90%
Enrichment category coverage Manual Confidence-guided
User feedback: "sounds like me" N/A >85%
Time to usable twin Hours <30 min
Trait confidence coverage 0% >80% across all dimensions

Related Features