RAG-Banking-Chatbot

The RAG-Banking-Chatbot is an AI-powered chatbot designed to provide personalized financial assistance by combining structured and unstructured data. It leverages Retrieval-Augmented Generation (RAG) to improve the accuracy of responses by fetching relevant financial policies and customer-specific details before generating an answer.

Features

Customer Data Retrieval – Uses an SQLite database to store and fetch structured financial data.

• Policy Document Search – Utilizes ChromaDB to store and retrieve unstructured financial policies.
• Retrieval-Augmented Generation (RAG) – Enhances chatbot responses by combining retrieved information with LLM-generated text.
• Retrieval-Augmented Generation (RAG) – Enhances chatbot responses by combining retrieved information with LLM-generated text.
• Personalized Banking Advice – Tailors responses based on customer details.
• Logging & Debugging – Tracks chatbot interactions and retrieval performance.
• Scalable Architecture – Modular design for easy integration and expansion.
• Unit & Integration Testing – Ensures reliability and correctness.
• Dockerized Deployment – Supports containerized execution for portability.
• Reinforcement Learning – Uses a feedback loop to improve the chatbot's performance.

Setup & Installation

python -m venv venv
source venv/bin/activate  # On Windows use: venv\Scripts\activate
pip install -r requirements.txt

Initialize the Database

python src/database/setup_database.py

#Index Loan Policies in ChromaDB

python src/retrieval/index_policies.py

Run the Chatbot

python src/chatbot/rag_chatbot.py

How It Works

1. User asks a financial question (e.g., "Am I eligible for a low-interest loan?")
2. Chatbot retrieves customer financial data from the SQLite database.
3. Relevant financial policies are fetched from ChromaDB.
4. A Retrieval-Augmented Prompt is generated, combining retrieved information with the user query.
5. LLM processes the prompt and generates a response.
6. The chatbot returns a factual, personalized answer.

Technologies Used

• Python (Primary programming language)
• SQLite (Structured data storage)
• ChromaDB (Vector database for unstructured data retrieval)
• LangChain (LLM and retrieval integration)
• OpenAI GPT API (Language model for response generation)
• Docker (Containerization for deployment)
• Logging & Testing Frameworks (Ensuring reliability and monitoring)

Future Enhancements

• Multi-turn conversation support
• Enhanced fact-checking for more reliable responses
• Real-time banking transaction analysis
• Web-based chatbot interface (FastAPI/Flask integration)