🧠 AI Wikipedia Speedrunner Pro

A smart, autonomous AI agent that plays the Wikipedia Game (navigating from a Start Page to a Target Page using only internal links).

This is the Pro Version of the speedrunner. Unlike basic bots, this version features Long-Term Memory (SQLite), Interactive Graph Visualization, and a modular architecture. It runs live on the internet without needing massive database dumps.

🔴 Live Demo: Try the AI Application

---

🚀 Pro Features

• 🧠 Semantic Brain: Uses all-MiniLM-L6-v2 to understand conceptual relationships (e.g., King is close to Queen).
• 💾 Long-Term Memory: Uses a local SQLite Database. If the AI solves a path once, it remembers it forever. The next time you ask for that path, it loads instantly!
• 🕸️ Interactive Graphs: Generates dynamic HTML network graphs (using Pyvis) to visualize the exact path the AI took.
• ⚡ Hybrid Scoring:
  • Vector Similarity: Finds conceptually related links.
  • Keyword Stacking: Rewards unique word matches (fixes the "Institute/University" trap).
  • Hub Strategy: Uses a "Panic Button" to route through global hubs (e.g., India, Science) if confused.
• 🐳 Dockerized: Fully containerized for easy deployment anywhere.

---

📂 Project Structure

The project has been refactored from a single script into a professional modular architecture:

Speedrun_The_Wifipidia/
│
├── src/
│   ├── ai_logic.py      # The Brain: Vector embedding & scoring math
│   ├── scraper.py       # The Hands: Live Wikipedia API interaction
│   ├── database.py      # The Memory: SQLite handling for saving runs
│   └── visualizer.py    # The Eyes: Network graph generation
│
├── app.py               # The Interface: Streamlit Dashboard
├── Dockerfile           # Production container configuration
├── requirements.txt     # Dependencies
└── README.md            # Documentation

---

🛠️ Installation & Usage

Method 1: Local Python Run

# 1. Clone the repo
git clone https://github.com/Akshat8510/Speedrun_The_Wifipidia.git
cd Speedrun_The_Wifipidia

# 2. Install dependencies
pip install -r requirements.txt

# 3. Run the app
streamlit run app.py

Method 2: Docker Run (Recommended for Deployment)

# 1. Build the image
docker build -t wiki-speedrunner .

# 2. Run the container
docker run -p 8501:8501 wiki-speedrunner

Access the app at http://localhost:8501.

---

🧠 The Logic (How it thinks)

The AI uses a Beam Search algorithm with a custom priority score:

$$ Score = S_{vector} + S_{keywords} + S_{hub} $$

1. Check Memory: First, it queries history.db. If the path exists, return it instantly.
2. Scan Links: If new, it fetches live links from the current Wikipedia page.
3. Evaluate:
   • Vector Score: Is the link conceptually close? (e.g., Physics -> Math)
   • Keyword Bonus: Does the link share unique words with the target?
   • Hub Rescue: If scores are low, jump to a Hub (e.g., United States) to reset context.
4. Visualize: Once the target is found, it draws the path using networkx and pyvis.

---

🧪 Example Benchmarks

Start Page	Target Page	Result	Logic Used
SpongeBob SquarePants	Nuclear Power	✅ Success	Vector Similarity
Achyuta	KIIT University	✅ Success	Hub Strategy (via "India")
Vector	Complex Number	✅ Success	Keyword Disambiguation
Pikachu	Adolf Hitler	✅ Success	Hub Strategy (via "Japan")

---

⚙️ Configuration

You can tweak the AI's "Personality" in src/ai_logic.py:

GLOBAL_HUBS = {"india", "united states", "science", ...} # Add more hubs here

Or in app.py:

BEAM_WIDTH = 3        # Higher = Smarter but Slower
MAX_STEPS = 50        # Max depth

---

📜 License

This project is open-source and available under the MIT License.

Built with ❤️ by Akshat8510.