DrLAW - AI Legal Assistant

Project Theory and Architecture

DrLAW is an innovative AI-powered legal assistance system designed to make legal information more accessible to the Indian population. The project combines several advanced technologies to provide accurate, context-aware legal advice in multiple Indian languages.

Core Components

1. Retrieval Augmented Generation (RAG) System

   • Uses a hybrid approach combining retrieval and generation
   • Indexes legal documents for accurate context retrieval
   • Employs semantic search for finding relevant legal information
   • Components:
     • Document Processor (PDF extraction)
     • Text Chunker (semantic splitting)
     • Embedding Model (sentence-transformers)
     • Vector Database (FAISS)
     • LLM Integration (Google Gemini)

2. Multilingual Support

   • Supports 11 Indian languages:
     • Hindi, Bengali, Telugu, Marathi, Tamil
     • Gujarati, Kannada, Malayalam, Punjabi, Odia
   • Uses Google Translate API for accurate translations
   • Maintains legal terminology accuracy

3. Knowledge Base

   • Processed legal documents including:
     • Indian Constitution
     • Major Acts and Laws
     • Legal Precedents
     • Legal Procedures
   • Organized in semantic chunks for efficient retrieval
   • Regularly updated with new legal information

Technical Implementation

1. Backend Architecture (Flask)

   /backend
   ├── app.py              # Main application logic
   ├── static/            # Static assets
   ├── templates/         # HTML templates
   ├── storage/           # Vector DB and chunks
   └── requirements.txt   # Dependencies
   

2. Frontend Structure

   /frontend
   ├── index.html        # Main interface
   ├── login.html        # Authentication
   ├── front.html        # Landing page
   └── config.js         # Configuration
   

3. Database Schema (Supabase)

   -- Users table
   CREATE TABLE users (
       user_id SERIAL PRIMARY KEY,
       username VARCHAR(255) NOT NULL,
       email VARCHAR(255) UNIQUE NOT NULL,
       password_hash VARCHAR(255),
       google_id VARCHAR(255) UNIQUE
   );
   
   -- Chats table
   CREATE TABLE chats (
       chat_id SERIAL PRIMARY KEY,
       user_id INTEGER NOT NULL REFERENCES users(user_id),
       question TEXT NOT NULL,
       answer TEXT NOT NULL,
       timestamp TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP
   );

RAG System Workflow

1. Document Processing

   • PDFs are processed and split into semantic chunks
   • Each chunk is embedded using sentence-transformers
   • Embeddings are stored in FAISS index
   • Metadata maintained for source tracking

2. Query Processing

   • User question is embedded
   • Similar chunks retrieved from FAISS
   • Context assembled from relevant chunks
   • Prompt constructed with legal format

3. Response Generation

   • Gemini API generates detailed response
   • Response structured with:
     • Legal analysis
     • Applicable laws
     • Required documentation
     • Step-by-step guidance

4. Translation Flow

   • Original response in English
   • Translation to requested language
   • Format preservation in translation

Setup and Configuration

Environment Variables

FLASK_SECRET_KEY=<secret>
GEMINI_API_KEY=<your-key>
SUPABASE_URL=<url>
SUPABASE_KEY=<key>
GOOGLE_CLIENT_ID=<id>
GOOGLE_CLIENT_SECRET=<secret>

Installation

1. Clone and setup:

git clone https://github.com/utknig123/DRLAW.git
cd DRLAW

2. Backend setup:

cd backend
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

3. Start server:

flask run

System Requirements

• Python 3.8+
• 4GB+ RAM (for embeddings)
• Storage for vector database
• Internet connection for APIs

Security Considerations

1. Authentication

   • Session-based auth
   • Google OAuth integration
   • Password hashing
   • CORS protection

2. Data Protection

   • Encrypted storage
   • Secure API calls
   • Rate limiting
   • Input sanitization

Future Enhancements

1. Technical Improvements

   • Real-time document updates
   • Improved semantic search
   • Caching system
   • Load balancing

2. Feature Additions

   • Document upload interface
   • Expert verification system
   • Legal form generation
   • Citation system

Troubleshooting

Common issues and solutions:

1. Missing API keys
2. Database connection errors
3. Memory issues with embeddings
4. Session management problems
5. CORS configuration issues

License

This project is licensed under the MIT License.