https://chatwithpdfimages.streamlit.app/
PDF Query Bot is an intelligent chatbot application designed to help users query and extract information from PDF documents. The bot leverages OpenAI's GPT-3.5 model and embedding techniques to provide accurate answers based on the content of the PDFs.
- PDF Text Extraction: Extracts text and images from PDF files.
- Text Chunking: Splits the extracted text into smaller snippets for better processing.
- Embeddings Creation: Generates embeddings for the text snippets using OpenAI's embedding model.
- Question Answering: Matches user queries with relevant snippets from the PDFs and generates a response using GPT-3.5.
- Streamlit Interface: Provides an interactive web interface for users to interact with the bot and visualize the results.
- Python 3.7+
- Streamlit
- pypdf
- langchain
- openai
- numpy
- fitz (PyMuPDF)
-
Clone the repository:
git clone https://github.com/yourusername/pdf-query-bot.git cd pdf-query-bot -
Install the required packages:
pip install -r requirements.txt
-
Set up your OpenAI API key. You can either:
- Replace the placeholder in the code with your API key, or
- Store the API key in an environment variable.
-
Run the application:
streamlit run app.py
- Upload PDF Files: Place your PDF files in the designated folder.
- Load PDFs: The bot extracts text and images from the PDFs.
- Create Embeddings: The text is split into snippets and embeddings are created.
- Query the Bot: Use the Streamlit interface to ask questions. The bot will provide answers based on the content of the PDFs and display relevant images if available.
- app.py: Main Streamlit application.
- app_utils.py: Utility functions for text extraction, embeddings creation, and question answering.
- requirements.txt: List of required packages.
Contributions are welcome! Please create an issue or submit a pull request for any improvements or bug fixes.