Financial Q&A RAG System

Overview

This project implements a Retrieval-Augmented Generation (RAG) system for answering financial questions based on data from 10-K filings. It combines ElasticSearch for efficient information retrieval with a local Language Model (LLM) to generate accurate and context-aware responses to financial queries.

Features

• Uses a dataset of 10,000 financial question-answer pairs derived from 10-K filings
• Implements a RAG architecture using ElasticSearch and Ollama
• Provides a user-friendly interface using Streamlit
• Fully dockerized for easy deployment and scalability

How It Works

1. Data Preparation: The system preprocesses the financial Q&A dataset and indexes it in ElasticSearch.
2. Query Processing: When a user submits a question, the system retrieves relevant context from ElasticSearch.
3. Answer Generation: The retrieved context is then passed to a local LLM (Qwen 0.5b) to generate a response.
4. User Interface: The generated answer is displayed to the user through a Streamlit web interface.

Prerequisites

• Docker
• Docker Compose

Setup and Running

1. Clone the repository:

   git clone https://github.com/srmadani/financial-qa-rag.git
   cd financial-qa-rag

2. Build and start the Docker containers:

   docker-compose up --build

3. Once all services are up, access the application at http://localhost:8501

Usage

1. Open your web browser and navigate to http://localhost:8501
2. Enter your financial question in the text input field
3. Click "Ask" to submit your query
4. The system will process your question and display the answer

Project Structure

• app.py: Main Streamlit application
• data_prep.py: Data preprocessing and ElasticSearch indexing
• Dockerfile: Instructions for building the Docker image
• docker-compose.yaml: Defines and configures the Docker services
• requirements.txt: Lists the Python dependencies
• wait-for-elasticsearch.sh: Ensures ElasticSearch is ready before starting the app

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

Acknowledgements

• Dataset source: Kaggle - Financial Q&A 10k
• Preprocessing credit: Kaggle Notebook
• RAG design credit: DataTalksClub/llm-zoomcamp

Disclaimer

This system is designed for educational and research purposes. Always consult with qualified financial professionals for actual financial advice.