nirasha-nelki/Chatbot-PDF-Reader

Chatbot PDF Reader

This application is a conversational AI assistant that can answer questions based on the contents of PDF documents. It utilizes LangChain, FAISS, Pinecone, and Streamlit to build a chat interface that interacts with users in real time.

Features

• Conversational AI: Ask questions related to the content of the uploaded PDF documents.
• Document Retrieval: Retrieve relevant information from large documents.
• Supports Pinecone and FAISS: Optionally use Pinecone for vector storage or FAISS for local vector storage.
• Streamlit UI: User-friendly interface to interact with the AI assistant.

Prerequisites

• Python 3.8 or later
• Virtual environment (optional but recommended)

Installation

1. Clone the repository:

   git clone https://github.com/nirasha-nelki/Chatbot-PDF-Reader.git
   cd chatbot-pdf-reader

2. Create and activate a virtual environment:

   python -m venv .venv
   .venv\Scripts\activate  

3. Install the required packages:

   pip install -r requirements.txt

4. Set up environment variables by creating a .env file in the root directory with the following content:

   PINECONE_API_KEY=your_pinecone_api_key
   INDEX=your_pinecone_index_name
   GROQ_API_KEY=your_groq_api_key

5. If you plan to use Pinecone, ensure that the pincecone_flag is set to True in backend.py. Otherwise, leave it as False to use FAISS.

Usage

1. Ingest Documents: Run the following command to ingest documents into Pinecone (if using Pinecone):

   python uploaddocs.py

2. Run the Streamlit App: Start the application by running:

   streamlit run chatui.py

3. Interact with the Chatbot: Open the provided URL in your web browser, and start asking questions based on the uploaded documents.

Application Structure

• chatui.py: The main Streamlit app file that sets up the chat UI and handles user interactions.
• backend.py: Contains the logic for interacting with the language model, managing chat history, and retrieving relevant document sections.
• uploaddocs.py: Handles document ingestion into Pinecone, splitting the document into manageable chunks, and storing them in the vector store.
• requirements.txt: List of Python packages required to run the application.
• doc/: A folder to store the documents (PDF files) that will be used by the chatbot.

Acknowledgements

This project uses LangChain, Streamlit, Pinecone, and FAISS.