Safe Scope : Parental Web Monitoring System with LLM-Based Content Detection

– Developed Safe Scope, a backend-driven parental web monitoring system that converts raw browsing activity into actionable, risk-aware alerts using intent-based analysis. Implemented secure APIs and a risk-scoring mechanism to analyze search behavior, reduce false positives, and provide meaningful insights instead of raw activity logs.

---

Key Features

• Parental Authentication System
  Register, log in, and securely manage multiple children under one account.

• Multi-Child Management
  Parents can monitor multiple children, each with separate activity logs.

• Real-Time Browsing Activity Logging
  Chrome extension captures page titles, URLs, timestamps, and risk assessments.

• Hybrid Risk Detection Engine

  • Layer 1: Fast keyword and lightweight ML filtering
  • Layer 2: LLM-based contextual reasoning for risk categorization
  • Layer 3: Age-aware evaluation to ensure developmental appropriateness
  • Outputs structured JSON: category, risk level, confidence score

• Explainability Agent

  • Generates human-readable explanations for flagged content
  • Example: “This page was blocked because it contains implicit sexual references unsafe for minors.”

• Modern Admin Dashboard

  • Live alerts, risk trends, category & language analytics
  • Displays explanations for flagged content
  • Built with React, Chart.js, and Django backend

• Privacy and Compliance

  • Anonymized browsing logs
  • Encrypted storage in MongoDB
  • Consent warnings for parents

---

📁 Project Structure

safeweb/
├── chrome_extension/        ← Chrome extension files
├── data/                    ← (For static data files if required)
├── monitor/                 ← Django app (core logic and views)
│   ├── migrations/
│   ├── static/
│   ├── templates/
│   ├── utils/
│   │   ├── risk_engine.py ← Hybrid + age-aware risk scoring
│   │   ├── explainability_agent.py← LLM explanations
│   │   ├── alert_engine.py
│   │   ├── data_preprocessor.py
│   │   ├── nsfw_detector.py
│   │   ├── predict_behaviour.py
│   │   └── query_analyzer.py
│   ├── admin.py
│   ├── apps.py
│   ├── forms.py
│   ├── models.py
│   ├── mongo_config.py
│   ├── tests.py
│   ├── urls.py
│   └── views.py
├── safeweb/                 ← Django project configuration
├── manage.py

2️⃣ Setup Python Environment

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

3️⃣ MongoDB Setup

• Install and start MongoDB locally or use a cloud service.

• Update your MongoDB URI in:

  monitor/mongo_config.py

Example:

MONGO_URI = "mongodb://localhost:27017/"
DB_NAME = "safewebguard_db"

4️⃣ Django Setup

python manage.py makemigrations
python manage.py migrate
python manage.py createsuperuser
python manage.py runserver

5️⃣ Chrome Extension Setup

• Open chrome://extensions/
• Enable Developer Mode
• Click Load Unpacked
• Select the chrome_extension/ folder
• Set child email through the extension popup

---

🔄 System Flow Diagram

Child Browsing → Chrome Extension → Content Ingestion Layer → Risk Engine (Keyword + ML + LLM)
→ Explainability Agent → MongoDB Storage → Admin Dashboard → Real-Time Alerts / Analytics

---

🖥️ Usage Guide

• Visit: http://127.0.0.1:8000/
• Register as a parent.
• Add child emails.
• Install Chrome Extension on your child's device.
• Monitor browsing logs in real-time through the dashboard.

---

🔒 Important Notes

• All predictions rely on external APIs integrated in monitor/utils/.
• Ensure backend URL in the Chrome extension matches your Django server URL.
• Works with MongoDB only; no SQL databases configured.

---

📄 Screenshots

---

✍️ Developed By

• Nagulapally Bhargavi - https://github.com/bhargavi852004