Ryuk Chatbot

Ryuk is an AI-powered chatbot built using Flask, Gemini AI, and Retrieval-Augmented Generation (RAG) techniques. It allows users to upload documents (PDF, DOCX, TXT) and ask questions based on the content of those documents, with the chatbot providing context-aware responses. The project features a sleek, futuristic UI with glowing animations, a responsive chat interface, and robust document processing capabilities.

Features

• Document Upload and Processing: Upload PDF, DOCX, or TXT files, which are processed into chunks and embedded using Sentence Transformers for context-aware responses.
• Context-Aware Chat: Ask questions based on uploaded documents, and the chatbot retrieves relevant context using cosine similarity and embeddings.
• Gemini AI Integration: Leverages Google's Gemini AI for generating natural language responses.
• Futuristic UI: A dark-themed interface with glowing stars, floating bubbles, and a minimal loading animation.
• Responsive Design: Works seamlessly across devices with a flexible layout.
• Clear Chat Functionality: Easily clear the chat history and uploaded documents.
• Error Handling: Robust logging and error handling for file processing and API calls.

Technologies Used

Backend

• Flask: Lightweight web framework.
• Google Generative AI (Gemini): For generating context-based responses.
• Sentence Transformers: all-MiniLM-L6-v2 for text embeddings.
• PyPDF2: For PDF text extraction.
• python-docx: For DOCX text extraction.
• NumPy and scikit-learn: For cosine similarity computations.
• python-dotenv: For managing environment variables.
• Logging: For error tracking and debugging.

Frontend

• HTML/CSS/JavaScript: Core UI and logic.
• Quicksand Font: From Google Fonts.
• Custom Animations: Glowing stars, bubbles, and three-dot wave loader.

Other Tools

• Git and GitHub: Version control and repository hosting.
• Python Virtual Environment: Dependency management.

How It Works

1. Document Processing and Embedding

• File Upload: PDF, DOCX, or TXT via /upload endpoint.

• Text Extraction:

  • PyPDF2 for PDFs.
  • python-docx for DOCX.
  • UTF-8 or Latin-1 reading for TXT.

• Chunking: Text is split into 1000-character chunks.

• Embedding: Chunks are embedded using all-MiniLM-L6-v2.

2. Context Retrieval

• Query Embedding: User question is embedded similarly.
• Cosine Similarity: Compared against chunk embeddings.
• Top 3 Chunks: Retrieved if similarity > 0.1.
• Summary Requests: Use full document.

3. Response Generation

• Prompt: Built using context + user query.
• Gemini AI: Response generated via Gemini API (gemini-1.5-flash).
• Markdown Response: Displayed in chat.

4. Frontend Interaction

• Chat Interface: Send questions via text input.
• Loading Animation: Three-dot wave.
• Dynamic Update: Scrolls and appends messages.
• Clear Chat: Resets chat and uploaded docs.

Screenshots

Setup Instructions

Clone the Repository

git clone https://github.com/VisvaV/Ryuk-Chatbot.git
cd Ryuk-Chatbot

Create Virtual Environment

python -m venv venv 
venv\Scripts\activate

Install Dependencies

pip install -r requirements.txt

Configure Environment Variables

echo "GEMINI_API_KEY=your_api_key_here" > .env

Replace your_api_key_here with your actual Gemini API key.

Run the Application

python app.py

Usage

Upload a Document

• Click the paperclip icon.
• Select a valid document (PDF, DOCX, or TXT).
• Confirmation appears upon success.

Ask a Question

• Enter question (e.g., "Summarize the document").
• Click Send or press Enter.

Clear Chat

• Click "Clear" to reset everything.

Implementation Details

Backend (app.py)

• Flask config with 16MB file limit.

• Routes:

  • / - Main page
  • /upload - Upload handler
  • /chat - Chat query handler
  • /clear - Clear chat/documents

• Document parsing and embedding logic.

Frontend (index.html, style.css, script.js)

• HTML layout with chat, input, and buttons.
• CSS dark theme with neon elements.
• JavaScript manages chat updates and AJAX requests.

Limitations

• In-Memory Storage: Embeddings lost on restart.
• File Size Limit: 16MB maximum.
• Gemini Dependency: Requires API key and internet.
• No Authentication: No user login or sessions.

Contributing

1. Fork the repository.
2. Create a branch: git checkout -b feature-name
3. Commit changes: git commit -m "Add feature-name"
4. Push: git push origin feature-name
5. Open a pull request with details.