📚 Document Chat Assistant

This repository contains a Streamlit-based Document Chat Assistant, enabling users to interact with and query content from various uploaded documents. Users can upload multiple text formats (PDF, TXT, CSV, Excel) and even a ZIP archive containing multiple documents. The application leverages LangChain, langchain-community extensions, langchain-openai, and OpenAI embeddings and chat models to provide semantic search and Q&A functionalities over the uploaded documents.

Key Features:

• Document Upload: Upload multiple documents or a ZIP file containing many files.
• Semantic Search & Q&A: Ask questions in natural language; get contextually relevant answers sourced from your documents.
• Summarization: Generate a global summary of all uploaded documents.
• Metadata Awareness: Display metadata (e.g., PDF page numbers, CSV row counts) for each source reference.
• Model Selection: Switch between OpenAI models (e.g., gpt-4, gpt-3.5-turbo).
• History Management: Clear conversation history, clear all documents, and download your chat history.

Language Interface:
The UI and messages are presented in French.

---

Installation

Clone the repository and navigate to the project directory:

git clone https://github.com/sanlamamba/Document-Talk.git
cd Document-Talk

Make sure you have Python 3.11+ installed. We recommend using a virtual environment:

python3.11 -m rag_talk rag_talk
source rag_talk/bin/activate

if you are using conda you can create a new environment using the following command

conda create -n rag_talk python=3.11
conda activate rag_talk

Install the required dependencies from requirements.txt:

pip install --upgrade pip
pip install -r requirements.txt

Required Packages:

• streamlit>=1.14.0
• langchain>=0.0.9
• langchain-community==0.3.18
• langchain-openai==0.3.6
• openai==1.63.2
• pandas==2.2.3
• PyPDF2==3.0.1

Export your OPEN AI Key in the environment

Get your OpenAI API Key, You can obtain an openAI key : -> (Open AI Api Keys)[https://platform.openai.com/api-keys]

If you wish to learn more about Open AI API keys you can read more about it here : (Open AI Quickstart)[https://platform.openai.com/docs/quickstart]

Set your OpenAI API key as an environment variable:

export OPENAI_API_KEY="sk-..."

---

How to Run

Once all dependencies are installed and OPENAI_API_KEY is set, run:

streamlit run app.py

You should see output similar to:

You can now view your Streamlit app in your browser.

  Local URL: http://localhost:8501
  Network URL: http://<your_machine_ip>:8501

Open the URL in your browser to access the Document Chat Assistant interface.

---

Usage

1. Upload Documents:
   In the sidebar, upload one or multiple files (PDF, TXT, CSV, Excel) or a ZIP file containing multiple documents.

2. Ask Questions:
   Type your questions in the main interface. The assistant will respond with an answer and provide a list of source documents (with metadata and excerpts) used to generate the response.

3. Summarize Documents:
   Use the "Résumer tous les documents" button in the sidebar to get a global summary.

4. Manage Documents & History:

   • Clear all uploaded documents using the "Effacer tous les documents" button in the sidebar (you will need to refresh the page afterward).
   • Clear chat history using the "Effacer l'historique de conversation" button in the sidebar.
   • Download your chat history for record-keeping.

---

Model Selection

Use the dropdown in the sidebar to choose between gpt-4 and gpt-3.5-turbo. This selection affects the LLM used for Q&A and summarization.

---

Notes

• The application is in French, but can handle documents in various languages.
• Ensure your OpenAI API key has appropriate billing and model access.
• If you encounter version warnings or deprecation warnings, consider upgrading the respective libraries.

---

License

This project is provided under an MIT License. See LICENSE for details.