Law-Aware AI Chatbot

A sophisticated Agentic RAG (Retrieval-Augmented Generation) system tailored for Indian legal research, capable of understanding complex legal queries, synthesizing case law, and providing actionable legal advice.

Features

• Intelligent Legal Research: Automated scraping of judgments from eCourts portal
• Multi-Modal Data Processing: Handles both online judgments and offline legal PDFs
• Advanced RAG Pipeline: Combines semantic search (ChromaDB) with knowledge graphs (Neo4j)
• Agentic Reasoning: Uses LangChain/LangGraph for complex query orchestration
• Legal Entity Recognition: Extracts judges, acts, precedents, and case relationships
• CAPTCHA Handling: Integrated EasyOCR for automated form solving

Tech Stack

• Automation & Scraping: Python, Selenium, undetected-chromedriver, EasyOCR
• Document Processing: PyMuPDF, pdfplumber, LangChain text splitters, spaCy NLP
• Databases: ChromaDB (Vector store), Neo4j (Knowledge Graph)
• AI & Orchestration: LangChain/LangGraph, Groq/OpenAI/Anthropic/Gemini APIs
• UI: Streamlit/Gradio chatbot interface, Flask web app with Web3 integration

Installation

1. Clone the repository

   git clone <repository-url>
   cd law-aware-ai

2. Install dependencies

   uv sync

3. Set up environment variables

   cp .env.example .env
   # Edit .env with your API keys and database credentials

4. Install additional NLP models (optional)

   python -m spacy download en_core_web_sm

Usage

Command Line Interface

# Scrape judgments from eCourts
python main.py scrape --case-number "CRL.A. 123/2023" --court "Delhi High Court" --max-pages 5

# Process PDF documents
python main.py process --pdf-file data/pdfs/sample.pdf
python main.py process --pdf-dir data/pdfs/

# Store processed documents in vector database
python main.py store --processed-dir data/processed/
python main.py store --json-file data/processed/sample.json

# Populate and query knowledge graph
python main.py graph --populate --processed-dir data/processed/
python main.py graph --query "cases by judge Justice Singh"

# Run intelligent legal research agent
python main.py agent --query "Explain Section 302 IPC interpretation in murder cases"
python main.py agent --query "Compare death penalty approaches in Indian courts" --verbose

# Start web interface with Mars chatbot
python main.py web
python main.py web --host 127.0.0.1 --port 8000 --debug

Demo Scripts

# Phase 1-2 Demo: Data processing and vector storage
python demo_phase2.py

# Phase 3 Demo: Knowledge graph construction
python demo_phase3.py

# Phase 4 Demo: Agentic RAG pipeline
python demo_phase4.py

Project Phases

✅ Phase 1: Infrastructure & Setup (Complete)

• Project structure with modular src/ layout
• uv-based dependency management
• Pydantic settings with environment variables
• CLI interface with subcommands

✅ Phase 2: Data Processing & Vector Storage (Complete)

• eCourts web scraping with CAPTCHA handling
• PDF text extraction and entity recognition
• Legal-aware text chunking for semantic search
• ChromaDB vector storage with metadata

✅ Phase 3: Knowledge Graph Construction (Complete)

• Neo4j graph database integration
• Legal relationship modeling (cases, judges, acts, precedents)
• Graph population pipeline from processed documents
• Natural language graph querying

✅ Phase 4: Agentic RAG Pipeline (Complete)

• Intelligent query routing (vector/graph/hybrid/complex)
• LangGraph orchestration for multi-step reasoning
• Legal entity extraction and context enrichment
• Comprehensive answer synthesis from multiple sources
• Confidence scoring and reasoning transparency

📋 Phase 5: User Interface & Deployment

• Streamlit/Gradio chatbot interface
• Conversation memory and context management
• Production deployment with monitoring

Scrape judgments from eCourts

python main.py scrape --case-number "CRL.A. 123/2023" --court "Delhi High Court"

Process PDF documents

python main.py process --pdf-dir data/judgments/

### Development

```bash
# Install dev dependencies
uv sync --dev

# Run tests
pytest

# Format code
black src/

Project Structure

law-aware-ai/
├── config/
│   ├── settings.py          # Configuration management
│   └── __init__.py
├── data/
│   ├── judgments/           # Raw judgment PDFs
│   ├── processed/           # Processed text data
│   ├── vector_db/           # ChromaDB storage
│   └── graph_db/            # Neo4j data
├── src/
│   ├── data_processing/
│   │   ├── ecourts_scraper.py    # eCourts portal scraper
│   │   └── pdf_processor.py      # PDF text extraction
│   ├── vector_store/        # ChromaDB integration
│   ├── graph_store/         # Neo4j knowledge graph
│   ├── agent/               # LangChain orchestration
│   └── ui/                  # Streamlit/Gradio interfaces
├── main.py                  # CLI entry point
├── pyproject.toml           # Project configuration
├── .env.example             # Environment template
└── README.md

Architecture

Data Flow

1. Data Acquisition: Scrape eCourts portal and process offline PDFs
2. Text Processing: Extract and clean legal text, identify sections
3. Entity Extraction: Use NLP to identify legal entities and relationships
4. Vector Storage: Chunk and embed text in ChromaDB for semantic search
5. Knowledge Graph: Store relationships in Neo4j for structural queries
6. Agent Orchestration: Route queries to appropriate data sources
7. Response Synthesis: Generate comprehensive legal answers

Agentic Router

• Factual Queries: "What is the judgment in case X?" → ChromaDB
• Relational Queries: "How does case X relate to case Y?" → Neo4j
• Complex Queries: Multi-source reasoning with LLM synthesis

Development Roadmap

• Phase 1: Data acquisition and processing ✅
• Phase 2: Vector storage and semantic search
• Phase 3: Knowledge graph construction
• Phase 4: Agentic RAG implementation
• Phase 5: UI development and deployment

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests
5. Submit a pull request

License

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

Disclaimer

This system is designed for legal research assistance and should not be considered a substitute for professional legal advice. Always consult qualified legal professionals for specific legal matters.