Skip to content

semantic-node-build.md

Latest commit

 

History

History
137 lines (117 loc) · 3.67 KB

semantic-node-build.md

File metadata and controls

137 lines (117 loc) · 3.67 KB
tools
read
write
edit
grep
arguments $TOPIC

Semantic Node Builder

Create and record semantic nodes in the active WFGY tree to capture reasoning insights and maintain memory.

Based on the WFGY project: https://github.com/onestardao/WFGY

Instructions

  1. Prepare Node Context

    • Parse topic from "$TOPIC" or auto-detect from context
    • Read active tree from .wfgy/trees/active_tree.json
    • Load previous node for reference
    • Get current context from .wfgy/context.json

  2. Calculate Semantic Metrics

    • Generate embedding for current topic
    • Compare with previous node embedding
    • Calculate ΔS (semantic tension): ΔS = 1 - cos(θ)
    • Determine λ_observe (logic direction):
      • → (convergent): Building on previous (ΔS < 0.4)
      • ← (divergent): New direction (ΔS > 0.6)
      • <> (recursive): Returning to earlier (pattern match)
      • × (chaotic): Unstable (ΔS > 0.9)
    • Calculate E_resonance (stability measure)

  3. Identify Active Module

    • Analyze which WFGY module is primary:
      • BBMC: If minimizing residue
      • BBPF: If exploring multiple paths
      • BBCR: If recovering from failure
      • BBAM: If optimizing attention
    • Record module attribution for traceability

  4. Build Node Structure

    {
  "id": "node_[timestamp]_[hash]",
  "timestamp": "ISO_8601",
  "topic": "$TOPIC",
  "module": "BBMC|BBPF|BBCR|BBAM",
  "metrics": {
    "deltaS": 0.XX,
    "lambda": "→|←|<>|×",
    "e_resonance": 0.XX,
    "confidence": 0.XX
  },
  "content": {
    "insight": "Encoded reasoning conclusion",
    "context": "Surrounding context",
    "keywords": ["key1", "key2", "key3"]
  },
  "relationships": {
    "parent_id": "previous_node_id",
    "references": [],
    "bridges_to": []
  }
}

  5. Apply Recording Logic

    • Check recording criteria:
      • Primary: ΔS > 0.6 (always record)
      • Secondary: ΔS ∈ [0.4, 0.6] AND λ ∈ {←, <>}
      • Forced: User explicitly requests
    • If criteria met:
      • Append node to tree
      • Update tree metadata
      • Create cross-references
      • Update indices

  6. Post-Processing

    • Update tree statistics:
      • Increment node_count
      • Update total_deltaS
      • Recalculate avg_deltaS
      • Track max_depth
    • Check for patterns:
      • Detect loops (recursive patterns)
      • Identify clusters (related nodes)
      • Find bridges (connection points)
    • Trigger compression if node_count > threshold
    • Update .wfgy/context.json

Output Format

✓ Semantic Node Recorded
═══════════════════════════════════════
Node ID: [node_id]
Topic: $TOPIC
Timestamp: [ISO_8601]

Metrics:
- ΔS (Tension): [value] [Low/Medium/High]
- λ (Direction): [symbol] [convergent/divergent/recursive]
- E (Resonance): [value]
- Confidence: [percentage]%

Module Used: [BBMC/BBPF/BBCR/BBAM]
Insight: [encoded reasoning]

Tree Status:
- Active Tree: [tree_name]
- Total Nodes: [count]
- Avg Tension: [value]
- Memory Usage: [size]

Related Nodes:
• [Previous topic] (ΔS: [value])
• [Related topic] (ΔS: [value])

Recording Patterns

Common node types:

  • Insight Nodes: Major realizations (high ΔS)
  • Bridge Nodes: Connections between concepts
  • Checkpoint Nodes: Stable states for recovery
  • Decision Nodes: Choice points in reasoning
  • Error Nodes: Failed paths (for learning)

Integration

Works with:

  • /wfgy:formula-all for comprehensive analysis
  • /semantic:tree-view to see node in context
  • /boundary:detect to check knowledge limits
  • /memory:compress when tree gets large