Business QA Chatbot

Overview

The Business QA Chatbot project provides a conversational interface for business-related queries using a Streamlit application. It leverages advanced NLP and machine learning techniques to deliver accurate responses based on data extracted from PDF documents.

Project Structure

• streamlit_app/: Contains the main Python files and Docker configuration for running the Streamlit app.
  • Dockerfile: Docker configuration for building and running the application.
  • main.py: The main Streamlit application script.
  • requirements.txt: Lists all the required Python packages for the application.
• Business_QA_Chatbot.pynb: Jupyter Notebook for interactive development and testing, recommended to use in Google Colab.
• streamlit_run/data/: Directory containing PDF files used to train the model.
• vectorize_documentation.py: Script to convert PDF data into a vector database using ChromaDB.

Installation

Manual Installation

1. Clone the Repository

   git clone https://github.com/your-username/business-qa-chatbot.git
   cd business-qa-chatbot

2. Create a Virtual Environment (Optional but Recommended)

   python -m venv venv
   source venv/bin/activate  # On Windows use venv\Scripts\activate

3. Install Dependencies

   pip install -r requirements.txt

4. Rename Configuration Files

   Rename sample_config.json to config.json and replace the placeholder "YOUR_API_KEYS" with your actual API key.

5. Run the Streamlit App

   Navigate to the streamlit_app/ directory and start the Streamlit application:

   `streamlit run main.py`

Docker Deployment

1. Build the Docker Image

   Navigate to the streamlit_app/ folder where the Dockerfile is located. Build the Docker image using:

   docker build -t business-qa-chatbot .

2. Run the Docker Container

   After building the image, run the Docker container with:

   docker run -p 8080:8080 business-qa-chatbot

3. Access the Application

   Open your web browser and navigate to http://localhost:8080 to access the Streamlit app.

Docker Image Handling

Pulling the Docker Image

To get the latest Docker image from Docker Hub, use:

docker pull gauravwankhede/business-qa-chatbot:latest

Building the Docker Image Locally

1. Navigate to the Directory Containing the Dockerfile

cd path/to/streamlit_app

2. Build the Docker Image

docker build -t business-qa-chatbot .

Running the Docker Container

1. Run the Docker Container

docker run -p 8080:8080 business-qa-chatbot

PDF Data

All PDF files used for training the model are stored in the streamlit_run/data/ folder. Ensure these files are correctly placed for the application to access.

Vectorization

The vectorize_documentation.py file converts PDF data into a vector database using:

• UnstructuredPDFLoader from langchain_community for loading PDF documents.
• RecursiveCharacterTextSplitter from langchain_text_splitters for splitting text.
• HuggingFaceEmbeddings from langchain_huggingface for generating embeddings.
• Chroma from langchain_chroma for managing the vector database.

Development

For interactive development and testing, use the Business_QA_Chatbot.pynb Jupyter Notebook. It is recommended to run this notebook in Google Colab.

Contact

For any questions or feedback, please contact pgywww@gmail.com.