Pakistan Legal Guide

answer queries related to Pakistani laws

Comment

Pakistan Legal Assistant 🇵🇰

About the Project

The Pakistan Legal Assistant is an AI-powered chatbot designed to answer queries related to Pakistani laws. It provides accurate, detailed, and multilingual responses in English and Urdu, making legal information accessible to everyone.

This project was developed as part of the ULEFUSA - APTECH Generative AI Training Hackathon (Jan-Feb 2025).

Features

  • Multilingual Support: Answers legal questions in both English and Urdu.
  • Context-Aware Responses: Retrieves relevant legal documents and provides structured answers.
  • User-Friendly Interface: Simple and intuitive chat interface powered by Gradio.
  • Fast and Accurate: Uses state-of-the-art AI models for quick and reliable responses.

How It Works

  1. User Query: The user asks a legal question in English or Urdu.
  2. Language Detection: The system detects the language of the query.
  3. Document Retrieval: Relevant legal documents are retrieved from a vector store.
  4. Response Generation: The system generates a detailed and structured response in the same language as the query.
  5. Output Delivery: The response is displayed in the chat interface.

Technologies Used

Here’s a breakdown of the technologies and models used in this project and their roles:

Technology/Model Role
Python The primary programming language used for the entire project.
Gradio Used to create the user-friendly chat interface.
FAISS Efficient vector store for document retrieval and similarity search.
Llama3-8b-8192 (Groq API) Large Language Model (LLM) used for generating detailed responses.
Helsinki-NLP/opus-mt-en-ur Translation model for converting English context into Urdu.
Multilingual MiniLM-L12-v2 Multilingual embeddings model for understanding and retrieving text.
LangChain Framework for retrieval-augmented generation (RAG) and prompt engineering.
Hugging Face Transformers Library for multilingual embeddings and translation models.
PyPDF2 Used to extract text from PDF documents.

Live Demo

Try the Pakistan Legal Assistant live:
👉 Live Demo on Hugging Face

Project Slides

Check out the project slides for a detailed overview:
📑 Slides on Canva

Demo Video

Watch the demo video to see the project in action:
🎥 Demo Video on Google Drive

Installation

To run this project locally, follow these steps:

  1. Clone the Repository:

    git clone https://github.com/Qasim-Gill/pakistan-legal-assistant.git
cd pakistan-legal-assistant

  2. Install Dependencies:

    pip install -r requirements.txt

  3. Set Up Environment Variables:

    • Create a .env file and add your Groq API key: plaintext GROQ_API_KEY=your_groq_api_key_here

  4. Run the Application:

    python app.py

  5. Access the App:

    • Open your browser and go to http://127.0.0.1:7860.

Usage

  1. Enter your legal question in the chat interface.
  2. The system will retrieve relevant legal documents and generate a detailed response.
  3. Ask follow-up questions or explore example queries provided.

Example Queries

  • English: "What is the punishment for theft?"
  • Urdu: "چوری کی سزا کیا ہے؟"
  • English: "Explain Section 302 of PPC."
  • Urdu: "PECA 2016 کے تحت سائبر کرائم کی سزائیں؟"

Contributors

  • Ali Rayan - Project Lead
  • Muhammad Qasim - Backend Developer
  • Yousuf - Backend Developer
  • Alisha Tariq - Slides and Video

License

This project is licensed under the MIT License. See the LICENSE file for details.

Acknowledgments

  • ULEFUSA - APTECH - ICodeGuru - ASPIRE PAKISTAN - UET - Pak Angeles for hosting the Generative AI Training Hackathon.
  • Hugging Face for providing the platform to deploy the app.
  • Groq for the fast and efficient language model API.

Contact

For questions or feedback, feel free to reach out:
📧 Email: mq77gill@gmail.com

Support the Project

If you find this project useful, please give it a ⭐ on GitHub!

Detailed Model Descriptions

Here’s a deeper dive into the models and their roles:

  1. Python:

    • The backbone of the project, used for scripting, logic, and integrating all components.

  2. Gradio:

    • Provides a simple and interactive chat interface for users to interact with the system.

  3. FAISS:

    • A highly efficient library for similarity search, used to retrieve relevant legal documents based on user queries.

  4. Llama3-8b-8192 (Groq API):

    • A powerful language model used to generate detailed and structured responses based on the retrieved context.

  5. Helsinki-NLP/opus-mt-en-ur:

    • A translation model specifically trained for English-to-Urdu translation. It translates the retrieved English context into Urdu for Urdu queries.

  6. Multilingual MiniLM-L12-v2:

    • A multilingual embeddings model used to convert text into numerical vectors, enabling the system to understand and retrieve text in multiple languages.

  7. LangChain:

    • A framework for building retrieval-augmented generation (RAG) pipelines. It handles document retrieval, prompt engineering, and response generation.

  8. Hugging Face Transformers:

    • A library that provides pre-trained models for embeddings and translation, used to power the multilingual capabilities of the system.

  9. PyPDF2:

    • A library for extracting text from PDF documents, used to process legal documents stored in PDF format.

Built With

Share this project:

Updates