TheOrbotAI/ai-contract-risk-analyzer

⚖️ AI Contract Risk Analyzer

An AI-powered web application for analyzing legal contracts using NLP, Machine Learning, and Large Language Models.

🧠 Overview

AI Contract Risk Analyzer is a complete AI-powered web application that helps users analyze legal contracts and extract key insights automatically. It uses Local LLM models + Gemini AI + NLP + Vector Search to process and understand legal documents intelligently — all running locally on your machine with no data leaving your device.

🚀 Live Demo

🌐 Deployed App: Click here to try it live

✨ Features

🛠️ Tech Stack

📁 Project Structure

ai-contract-risk-analyzer/ │ ├── app.py ← Main Streamlit application ├── requirements.txt ← All dependencies ├── README.md ← Project documentation │ ├── utils/ │ ├── pdfloader.py ← PDF text extraction │ ├── chunker.py ← Text splitting into chunks │ ├── embeddings.py ← Sentence embeddings │ ├── vectorstore.py ← FAISS vector index │ ├── retriever.py ← Semantic retrieval │ ├── llm.py ← Local Flan-T5 model │ ├── riskanalyzer.py ← Risk scoring & keywords │ ├── summarizer.py ← BART summarization │ ├── nerextractor.py ← Named entity extraction │ ├── entitygraph.py ← PyVis network graph │ ├── geminillm.py ← Gemini AI integration │ ├── translator.py ← Urdu translation │ └── pdfreport.py ← PDF report generation

⚙️ Installation & Setup

1. Clone the repository: git clone https://github.com/Aleesha-Aslam/ai-contract-risk-analyzer.git cd ai-contract-risk-analyzer

2. Install dependencies: pip install -r requirements.txt

3. Run the application: streamlit run app.py

4. Open browser: http://localhost:8501

🔑 Gemini AI Setup (Optional)

1. 1.Go to → https://aistudio.google.com
2. 2.Click "Get API Key" → Create new key
3. 3.In the app sidebar → Toggle "Use Gemini AI" → Paste your key
4. 4.Enjoy much better AI responses!

🚨 Risk Analysis System

The system detects risky legal patterns using weighted keyword matching:

5 Category Scores:

• —💰 Payment Terms — Are payment conditions risky?
• —⚖️ Liability — Are liability clauses vague?
• —📅 Termination — Can contract be terminated unfairly?
• —🔒 Confidentiality — Is data protection adequate?
• —📋 Obligations — Are obligations clearly defined?

🤖 AI Models Used

📊 How It Works

1. 1.User uploads PDF contract
2. 2.Text extracted using PyMuPDF
3. 3.Text split into chunks (500 chars each)
4. 4.Embeddings created using Sentence Transformers
5. 5.FAISS indexes all chunks for fast search
6. 6.User asks a question
7. 7.Top relevant chunks retrieved semantically
8. 8.LLM generates answer from retrieved context
9. 9.Risk analyzer scans for 30+ risky keywords
10. 10.Risk score (0-100) calculated with categories
11. 11.Professional PDF report generated

🌍 Urdu Translation Support

The app supports Urdu translation of:

• —📄 Contract Summary
• —🚨 Risk Analysis Report
• —📝 Full Contract Text

Using Google Translate via deep-translator library with proper RTL (Right-to-Left) text rendering.

🎨 UI Features

• —Modern Purple Theme with dark mode
• —Floating Particles animated background
• —Wave Animation at bottom of screen
• —Card Hover Glow effect
• —Progress Bar smooth fill animation
• —Badge Pop-In bounce animation
• —Pulsing Buttons with glow effect
• —Custom Purple scrollbar

📋 Requirements

streamlit pymupdf sentence-transformers faiss-cpu transformers torch numpy reportlab deep-translator pyvis google-genai beautifulsoup4

🎓 About This Project

This project was developed as an AI Course Assignment demonstrating:

• —Natural Language Processing (NLP)
• —Machine Learning and Deep Learning Models
• —Large Language Models (LLM)
• —Information Retrieval using FAISS Vector Search
• —Full Stack Web Application using Streamlit
• —API Integration with Gemini AI
• —Document Processing with PDF
• —Data Visualization using PyVis Graphs
• —Urdu Language Support

👩‍💻 Author

Aleesha Aslam GitHub: https://github.com/Aleesha-Aslam

📜 License

This project is developed for educational purposes as part of an AI course assignment.

Made with ❤️ by Aleesha Aslam ⚖️ AI Contract Risk Analyzer — v3.0 Final