RAG-based Persona Chatbot 🤖

A Retrieval-Augmented Generation system that learns and mimics personal communication styles from chat history.

🏗 System Architecture

graph TD
    subgraph Data Ingestion
        CH[Chat History] --> Parser[Chat Parser]
        Parser --> QA[Q&A Pairs]
        QA --> VE[Vector Embeddings]
    end

    subgraph Vector Storage
        VE --> VS[(Chroma DB)]
        VS --> IC[Index Collection]
        IC --> MT[Metadata Tags]
    end

    subgraph Persona Analysis
        QA --> PA[Personality Analyzer]
        PA --> PT[Traits Extraction]
        PA --> PS[Style Analysis]
        PA --> PP[Phrase Patterns]
    end

    subgraph RAG Engine
        UQ[User Query] --> SR[Semantic Retrieval]
        SR --> VS
        VS --> RC[Relevant Context]
        RC --> RG[Response Generator]
        PT --> RG
        PS --> RG
        PP --> RG
    end

    subgraph Response Pipeline
        RG --> RM[Response Matching]
        RM --> EX[Exact Match]
        RM --> SM[Similar Match]
        EX --> FR[Final Response]
        SM --> FR
    end

Loading

✨ Features

🧠 Intelligent Persona Analysis

• Extracts personality traits
• Learns communication style
• Identifies common phrases
• Maps topic interests

🎯 Context-Aware Responses

• RAG-powered generation
• Historical context matching
• Style-consistent replies
• Exact response matching

🚀 Quick Start

Prerequisites

python -m venv venv
.\venv\Scripts\activate
pip install -r requirements.txt

Setup

1. Set OpenAI API key

set OPENAI_API_KEY=your-api-key

2. Run application

python app.py

3. Access interface

• Open browser to http://localhost:5000
• Upload chat history file
• Start conversation

💻 Technologies

• Frontend: Bootstrap 5, Socket.IO
• Backend: Flask, Flask-SocketIO
• AI/ML: LangChain, OpenAI GPT-4
• Vector Store: ChromaDB
• Embeddings: OpenAI Ada

📁 Project Structure

Open_Source_AI_Hackathon/
├── app.py              # Flask server
├── persona_chatbot.py  # Core RAG logic
├── templates/          # Frontend
│   └── index.html     # Chat interface
├── requirements.txt    # Dependencies
└── README.md          # Documentation

📄 License

MIT License - See LICENSE for details

👥 Team

Developed for Open Source AI Hackathon 2024 by Data Doppelgangers (GitHub: @ivanye2509, @prabhatmenon, @patelmanan96, @solkit70)

🙏 Acknowledgments

• OpenAI
• LangChain Framework
• Flask Community
• Kaggle CommunityGit