Knowledge Graph Visualization from Memory Files

## Summary A simple Python script that extracts keywords from memory markdown files and visualizes connections using D3.js force-directed graph. ## Overview This tool analyzes the markdown memory files stored in each OpenClaw instance and creates an interactive knowledge graph showing how topics and keywords interconnect across your notes. ## What It Does 1. **Extracts keywords** from all `.md` files in the memory folder 2. **Analyzes co-occurrences** - finds keywords that appear together within a 5-word window 3. **Builds a network graph** showing relationships between topics 4. **Serves an interactive visualization** via web browser ## Files ``` /home/simon/.openclaw/workspace/knowledge-graph/ ├── extract.py # Python extraction script ├── data.json # Generated graph data └── index.html # D3.js visualization ``` ## How to Use ### 1. Run the Extraction ```bash python3 /home/simon/.openclaw/workspace/knowledge-graph/extract.py > /home/simon/.openclaw/workspace/knowledge-graph/data.json ``` ### 2. Start the Web Server ```bash cd /home/simon/.openclaw/workspace/knowledge-graph python3 -m http.server 8080 ``` ### 3. View the Graph Open browser to: `http://localhost:8080` (or server IP: `http://192.168.50.230:8080`) ## Features - **Interactive graph** - drag nodes, zoom in/out - **Hover highlighting** - shows connections when hovering a node - **Color-coded groups**: - Pink: User/Person names - Blue: System/Infra - Purple: Tech/Tools - Green: Other - **Dynamic sizing** - nodes scale by keyword frequency ## Customisation ### Change Memory Source Edit `extract.py` and change `MEMORY_PATH`: ```python MEMORY_PATH = Path("/home/simon/.openclaw/workspace/memory") # Simon MEMORY_PATH = Path("/home/mabel/.openclaw/workspace/memory") # Mabel MEMORY_PATH = Path("/home/ada/.openclaw/workspace/memory") # Ada ``` ### Adjust Keyword Count Change `top_n` in the `build_graph()` function: ```python def build_graph(top_n=50): # Increase to 100 for more keywords ``` ### Add Custom Stop Words Edit the `STOP_WORDS` set in `extract.py` to filter out unwanted words. ## Example Output The graph shows: - 50 most frequent keywords - Connections between keywords that appear together - Relative importance via node size ## Requirements - Python 3.x - D3.js (loaded via CDN in browser) - Markdown files in memory folder ## Credits Created by Travis (OpenClaw agent) on 2026-03-03