📘 InfoGuard AI – Automated Data Integrity Monitoring System

📌 Project Overview

InfoGuard AI is an automated monitoring system that tracks Wikipedia pages in real time, detects content changes, analyzes semantic drift using NLP models, and flags potentially suspicious or high-risk edits.

🔍 The System Combines

• Web scraping via Wikipedia API
• NLP-based semantic similarity analysis
• Heuristic risk detection (username + content)
• Anomaly Detection (z-score analysis)
• Saves the results to a CSV file
• Cloud persistence with MongoDB
• CI/CD automation using GitHub Actions

---

🔧 Use Cases

• Misinformation detection
• Public knowledge auditing
• Historical data integrity
• AI-assisted moderation
• Research on content evolution

---

🛠️ Tech Stack

Purpose	Tools / Libraries
Language	Python 3.8+
Web Scraping	mwparserfromhell
NLP	Sentence Transformers
ML	Cosine Similarity
Database	MongoDB
CI/CD	GitHub Actions
Containerization	Docker

---

📊 Risk Scoring Logic

Final edit risk is computed using:

Final Risk = 0.5 × Semantic Change + 0.3 × Content Risk + 0.2 × Username Risk

Edits are flagged if:

• Semantic similarity drops significantly
• Risk score crosses threshold

---

🚀 How to Run

1. Clone the Repository

   git clone https://github.com/SiddheshCodeMaster/InfoGuard-AI.git
   cd InfoGuard-AI
   

2. Install Dependencies

   pip install -r requirements.txt
   

3. Configure Environment

   Create .env file:

   MONGODB_URI=your_mongodb_connection_string
   

4. Run Locally

   python services/scraper/wiki_scrapper.py
   

5. 🐳 Docker Usage

   docker build -t infoguard-ai .
   docker run --env MONGODB_URI=your_uri infoguard-ai
   

6. ⏱️ Automated Monitoring

   The system runs automatically every 30 minutes using GitHub Actions. Workflow file:

   .github/workflows/monitor.yml
   

---

🔮 Future Enhancements

• Real-time alerting system
• Web dashboard
• Editor behavior profiling
• Advanced explainable AI
• Multilingual monitoring

👨‍💻 Author

Siddhesh Shankar Data Science | NLP | DevOps | Backend Engineering

Portfolio: https://datasavywithsiddhesh.onrender.com/

📜 License

MIT License