Legal AI Assistant & Summarizer

This repository contains two powerful AI-based tools designed for legal assistance and judgment summarization:

1. Legal Chatbot – A retrieval-based assistant that answers legal questions using Indian law sources.
2. Legal Summarizer – A fine-tuned BART model that compresses lengthy legal judgments into concise summaries.

---

Project Structure

LegalCompanion/
│
├── Legal_Chatbot/
│   ├── legal_chatbot_final.ipynb       # Main chatbot notebook
│   ├── chatbot_utils.py                # Helper functions for chatbot
│   └── vector_store/                   # FAISS index and ingested docs
│
├── Legal_Summarizer/
│   ├── bartseq2seq.ipynb               # BART fine-tuning notebook
│   ├── Legal_summarizer_final.ipynb    # Final summarization pipeline
│   └── summarizer_utils.py             # Supporting utilities
│
├── requirements.txt                    # Dependencies
├── LICENSE                             # MIT License
└── README.md                           # Project documentation

---

Features

Legal Chatbot

• Retrieval-Augmented Generation (RAG) using FAISS
• Ingests documents from Indian law: Constitution, IPC, CrPC, etc.
• Answers domain-specific legal questions
• Document-aware legal Q&A (PDF ingestion supported)

Legal Summarizer

• Fine-tuned BART model for legal judgment summarization
• Combines extractive and abstractive techniques
• Trained on d0r1h/ILC

---

Installation

1. Clone the repository:

git clone https://github.com/Anshul-ydv/Legal-ai.git
cd LegalCompanion

2. Install dependencies:

pip install -r requirements.txt

---

Usage

Run the Legal Chatbot

1. Navigate to the Legal_Chatbot/ folder.
2. Open and run:

legal_chatbot_final.ipynb

3. Follow the steps to load your corpus, generate embeddings, and interact with the assistant.

Run the Legal Summarizer

1. Go to Legal_Summarizer/.
2. Open and run:

Legal_summarizer_final.ipynb

3. Provide legal text input and get summarized outputs using the fine-tuned model.

---

Dataset Info

• Summarizer Dataset: d0r1h/ILC
• Chatbot Corpus: QA pairs and legal documents from Indian legal sources (manually or automatically curated)

---

Tech Stack

• Python · PyTorch · HuggingFace Transformers
• FAISS · Google Colab · NLTK · Sklearn
• Optional UI: Streamlit or Gradio

---

Future Improvements

• Add multilingual query support (Hindi, Bengali, etc.)
• Deploy as a web app using FastAPI + Streamlit
• Add support for OCR and audio-based legal queries
• Visual analytics dashboard for legal search trends

---

Acknowledgements

• Indian Legal Corpus – d0r1h/ILC
• HuggingFace Ecosystem (Transformers, Datasets)
• FAISS by Meta AI
• OpenAI APIs for comparison and benchmarking

---

License

This project is open-sourced under the MIT License.

---

Fun Fact

“This AI has read more Indian laws than most lawyers.”