Customer Assistant RAG Chatbot

This project implements a Retrieval-Augmented Generation (RAG) pipeline for building a customer support chatbot.

---

Features

• Preprocessing and Understanding:
  • Includes data understanding, parsing and processing strategies for feature creation and initializing vector databases.
• Retrieval-Augmented Generation (RAG):
  • Combines document retrieval with LLM-based response generation.
• Customizable Prompts:
  • Uses system, retriever, and user prompts for better control over the chatbot's behavior.
• Evaluation Tools:
  • Includes scripts for dataset validation, evaluation metrics, and pipeline testing.
• Streamlit UI:
  • A user-friendly interface for interacting with the chatbot.
• FastAPI Backend:
  • Provides an API for generating responses programmatically.

---

Prerequisites

1. Python Version:
   • Ensure you have Python 3.12 installed.
2. Install Dependencies:
   • Use pip to install the required dependencies from requirements.txt.
3. Set Python Path:
   • Setting the Python path is crucial for importing packages correctly.

---

Setting Up the Application

1. Clone the Repository

git clone https://github.com/your-repo/CustomerSupportAssistant_RAG_Challenge.git
cd CustomerSupportAssistant_RAG_Challenge

2. Create a Virtual Environment

python3 -m venv venv
source venv/bin/activate 

3. Install Dependencies

pip install --upgrade pip
pip install -r [requirements.txt]

4. Set the Python Path

To ensure that all modules are imported correctly, set the PYTHONPATH environment variable:

For mac and linux

export PYTHONPATH=$(pwd)

For Windows:

set PYTHONPATH=%cd%

Running the Application

1. Start the Backend (FastAPI)

Run the FastAPI backend to expose the chatbot API:

# python src/backend/api/app.py
uvicorn src.backend.api.app:app --host 0.0.0.0 --port 8085

The API will be available at http://localhost:8085.

2. Start the Streamlit UI

Run the Streamlit application for a user-friendly interface:

streamlit run src/ui/streamlit_app.py 

The UI will be available at http://localhost:8501.

Important Notes

• Setting Python Path:

Always set the PYTHONPATH environment variable before running the application to avoid import errors.

• Docker Support:

A Dockerfile is included for containerizing the application. Future work includes adding docker-compose for connecting the UI and backend.

Future Work

• Add Docker Compose for seamless integration of UI and backend.
• Implement evaluation metrics for generation and retrieval using RAGAS.
• Enhance the chatbot with Chain-of-Thought reasoning for better explanations.
• Expand support for multilingual queries.
• Add test cases for each functionalities.

Limitations

• Currently supports only English queries.
• Requires a pre-configured vectorstore for document retrieval.

Additional Resources

some interesting repositories

• Awesome LLM Interpretability
• Prompting Guide
• System Prompts and Models (hidden system prompts from cursor and more systems)
• Camel AI