This project provides an interactive research assistant designed to enhance your document comprehension and navigation. The assistant allows users to query multiple PDFs for specific answers, retrieving relevant document sections. It includes an intuitive interface for viewing and navigating to specific pages of PDFs or PowerPoint files, offering targeted previews for improved understanding.

Getting Started

Prerequisites

Ensure you have all required dependencies installed. You can do this by running:

pip install -r requirements.txt

Setup

1. Place the vector_store directory (for storing vectorized document representations) and the nlp_data directory (containing your PDFs) in the project's root directory.
2. Ensure all necessary files are available in these directories.

Running the Application

1. Start the Flask API:

   python backend.py

   This will launch the API on localhost:5003.

2. Run the Streamlit application:

   streamlit run app.py

   The interface will be available at localhost:8501 on your local machine.

3. Optionally, explore the demo.py notebook for additional functionalities and demonstrations of the project's capabilities.

RAG Evaluation

1. Create evaluation dataset using the code provided in prepare_ragas_set.ipynb.

2. Run the code in run_ragas.py.

3. The evaluation results will be saved in mini_result.txt.