Between two concepts, find the third that doesn't yet exist.
GAT-Powered Semantic Network Exploration - From fuzzy intuition to precise discovery
WordCloud Emergence is a creative tool and a way of thinking.
It won't tell you "dragon + girl = what," but unfolds a fuzzy territory where concepts converge: Medusa, Xiaolongnü, Maleficent, Jade Fox...
At the intersections of these neighbors, you'll find the answer that belongs only to you.
Scene One: The Classroom
A student asks: "What is justice?"
A good teacher doesn't answer directly. She says: "There's a story about a watchman..."
The student begins to explore: guard duty, responsibility, solitude, promise... Among these fuzzy neighbors, they walk toward their own understanding of justice.
Scene Two: The Creative Studio
Versace, Year of the Dragon campaign.
The creative director won't simply design "a dragon with a girl."
They write on the whiteboard: Dragon · Girl
Then ask: "In all of human cultural memory, who lives in both words?"
The team begins listing fuzzy neighbors:
- Medusa — Snake-haired gorgon, echoing dragon's reptilian form
- Xiaolongnü — Cool serenity and mysterious power in Eastern wuxia
- Maleficent — Dark queen transforming into dragon in Sleeping Beauty
- Jade Fox — Rebellious soul and weapon in Crouching Tiger, Hidden Dragon
- The Girl with the Dragon Tattoo — Violent aesthetics and revenge in Nordic noir
- Chihiro — Girl growing before the white dragon god in Spirited Away
- Nezha — Gender-fluid dragon prince embodying rebellion and rebirth
The room goes quiet.
Creativity doesn't lie in the direct stacking of "dragon + girl," but at the intersections of these neighbors.
Fuzziness Guides Precision - We're not doing simple keyword matching. Through AI's "fuzzy understanding" capability, we guide users to discover precise connections between concepts. Just like human thinking: starting from fuzzy intuition, gradually crystallizing into concrete conceptual networks.
The best answers don't lie within concepts themselves.
They exist in the spaces between, where fuzzy neighbors dwell.
In teaching, this is Socratic questioning—stories that spark deeper thought.
In creativity, this is concept topology—surprises found at intersections.
"Dragon + Girl" stacked together is dull.
But create distance, and something else appears between.
Medusa brings serpentine danger. Xiaolongnü brings Eastern serenity. Maleficent brings the possibility of transformation. Jade Fox brings rebellious tension.
Students reach their own understanding here. Creatives find unexpected connections here. Writers see images that don't yet exist here.
This is emergence.
Step 1: Extract Core Concepts
Dragon + Girl
Step 2: Expand Neighbors in Knowledge Graph
What connects both dragon and feminine?
- Medusa — Gorgon with serpentine hair, echoing reptilian forms
- Xiaolongnü — Cool serenity and mysterious power in Eastern wuxia
- Maleficent — Dark queen transforming into dragon in Sleeping Beauty
- Jade Fox — Rebellious soul and weapon in Crouching Tiger, Hidden Dragon
- The Girl with the Dragon Tattoo — Violent aesthetics and revenge in Nordic noir
- Chihiro — Girl growing before the white dragon god
- Nezha — Gender-fluid dragon prince, rebellion and rebirth
Step 3: Find Unique Creativity at Intersections
These fuzzy neighbors are where true inspiration lives.
- AI receives user's fuzzy concepts
- Deep learning models understand semantic boundaries
- Generate multiple possible semantic directions
- Calculate attention weights for each related concept
- Weight = Semantic Similarity × Concept Importance
- Form weighted conceptual networks
- User clicks trigger new semantic expansion
- System records complete semantic paths
- From fuzzy starting point to precise destination
- Framework: FastAPI (inspired by Creative-Writing migration)
- AI/ML: OpenAI API, PyTorch (GAT Networks)
- Async: asyncio for concurrent processing
- APIs: RESTful endpoints for concept expansion
- Visualization: D3.js for force-directed graphs and word clouds
- UI: Vanilla JS with modern CSS
- Internationalization: i18n support for multiple languages
- Responsive: Mobile-friendly design
- Deployment: Vercel (serverless)
- Containerization: Docker support (from Creative-Writing)
- Monitoring: Basic logging and error handling
- Data Storage: JSON-based concept databases
- API Architecture: FastAPI patterns for scalable backends
- Visualization Techniques: Advanced D3.js implementations for cognitive state displays (adapted for word cloud weights)
- Deployment Configurations: Docker and Kubernetes setups for production
- Knowledge Graph Concepts: Graph-based semantic expansion algorithms
- 🧠 Fuzzy Semantic Understanding - AI's "fuzzy understanding," not exact matching
- ⚡ Attention Weights - Concept importance visualized through weights
- 🔄 Path Tracking - Record complete journey from fuzzy to precise
- 📊 Dynamic Layout - Weight-based force-directed graph layout
- 🎯 Spotlight Mode - Focus on current exploration path
Input: Space-Time
Fuzzy Understanding: time, space, dimension, relativity, travel, parallel universe, quantum, eternity
Attention Weight Analysis:
- Relativity (0.95) - Core physics theory
- Dimension (0.90) - Spatial concept
- Travel (0.85) - Sci-fi imagination
- Parallel Universe (0.82) - Theoretical physics
- Quantum (0.80) - Microscopic world
Precise Path Discovery:
Space-Time → Relativity → Einstein → E=mc²
Space-Time → Dimension → 4D Space → Spacetime Curvature
wordcloud2tester/
├── backend/ # Python FastAPI backend
│ ├── path-expand.py # Concept expansion logic
│ ├── test.py # API tests
│ └── requirements.txt
├── frontend/ # Web frontend
│ ├── index.html # Main page
│ ├── src/ # JavaScript sources
│ └── i18n-translations.js
├── config/ # Configuration files
│ └── vercel.json # Vercel deployment config
├── data/ # Data files and assets
├── docs/ # Documentation
│ ├── README.md # Detailed docs
│ └── DEPLOY_GUIDE.md
└── README.md # This file
- Python 3.8+
- Node.js 16+ (for development)
- OpenAI API key
-
Clone the repository
git clone https://github.com/hazelian0619/wordcloud2tester.git cd wordcloud2tester -
Setup backend
cd backend pip install -r requirements.txt # Set your OpenAI API key export OPENAI_API_KEY="your-key-here"
-
Run locally
# Start backend python path-expand.py # Open frontend in browser cd ../frontend python -m http.server 8000 # Visit http://localhost:8000
- Vercel: Automatic deployment from main branch
- Docker: See deployment guide in
docs/DEPLOY_GUIDE.md
This project incorporates elements from the Creative-Writing repository (https://github.com/hazelian0619/Creative-Writing) to enhance functionality, particularly in API architecture, visualization techniques, and deployment practices.
Contributions are welcome! Please feel free to submit a Pull Request.
MIT License - see LICENSE file for details