Python Question-Answering System with LangChain and Google Gemini

This project implements a document-based question-answering system using LangChain, Google Gemini, and Chroma. The system allows users to ask questions about a set of documents and receive accurate and relevant answers.

Project Structure

The project is divided into two main parts:

1. Database Generation (Jupyter Notebook):

   • A Jupyter notebook (qa_application.ipynb) is responsible for loading, processing, and indexing the source documents.
   • Documents are loaded from the file system.
   • The langchain_community.document_loaders.generic library provides the functionality to load documents.
   • Documents are divided into smaller chunks to facilitate processing.
   • Embeddings are generated for each chunk using the GoogleGenerativeAIEmbeddings model.
   • Embeddings and text chunks are stored in a vector database using the Chroma library.

2. Web Application (FastAPI):

   • A FastAPI server exposes a WebSocket endpoint for user interaction.
   • Users can connect to the server via WebSocket and submit questions.
   • Received messages are processed by the QAApplication class.
   • The QAApplication class loads the Chroma database and initializes the information retrieval chain.
   • The retrieval chain uses the Google Gemini language model (gemini-1.5-flash) to understand questions and retrieve relevant information from the database.
   • Answers are sent back to the user via WebSocket.

Main Files

• notebooks/qa_application.ipynb: Jupyter notebook for generating the Chroma database.
• src/main.py: Main FastAPI application file.
• src/utils/: Directory containing auxiliary modules.
  • qa_application.py: Defines the QAApplication class.
  • index.py: Contains the function to load the user interface HTML.
• src/data/chroma.sqlite3: The Chroma database file.
• .env: File to store environment variables (e.g., API key).

How to Run

1. Generate the Database:

   • Navigate to the notebooks directory in the terminal: cd notebooks
   • Open and run the qa_application.ipynb Jupyter notebook to create the Chroma database.
   • Ensure to specify the correct path to the source documents within the notebook.

2. Set Environment Variables:

   • Create a .env file at the root of the project (parent directory of src).
   • Define the following environment variables in the .env file:
     • GOOGLE_API_KEY: Your Google Cloud API key.
     • PERSIST_DIRECTORY: Path to the directory where the Chroma database will be stored. Make sure this path is accessible from within the src directory.

3. Install Dependencies:

   • Ensure you are in the project's root directory.

   • Execute the following command to install the dependencies:

     pip install -r requirements.txt

4. Run the Application:

   • Make sure you are in the project's root directory.

   • Run the following command to start the FastAPI server:

     uvicorn src.main:app --reload

5. Access the Application:

   • Open a web browser and go to http://127.0.0.1:8000/.

How to Use the Application

1. Access the Interface: Open your web browser and navigate to http://127.0.0.1:8000/. You will see a simple chat interface.
2. Type your Question: Type your question in the text box at the bottom of the page and click "Send".
3. View the Answer: The system's response will be displayed in the chat area above the text box.

Notes

• The GoogleGenerativeAIEmbeddings model requires a Google Cloud API key.
• Application performance depends on database size and machine processing power.
• The code includes comments explaining key steps and functionalities.

Future Improvements

• Implement authentication and authorization to restrict access to the application.
• Allow users to upload their own documents.
• Add support for different file types (e.g., PDF, Word).
• Improve the user interface.