🛡️ EduShield

AI-Powered Plagiarism & Authorship Detection System

A full-stack academic integrity platform built as a BSc IT Final Year Project

---

📖 Overview

EduShield is a comprehensive academic integrity system that detects plagiarism and AI-generated content in student documents. It supports two distinct workflows — a classroom-based Faculty–Student pipeline for institutional use, and a Personal Document Checker for individual users — making it equally useful for educators and independent researchers.

---

✨ Features at a Glance

Feature	Student	Faculty	Personal User
Upload Documents	✅	✅	✅
Plagiarism Detection	✅	✅	✅
AI Content Detection	✅	✅	✅
Downloadable PDF Report	—	—	✅
View Assignment Feedback	✅	—	—
Manage Sections	—	✅	—
Review Submissions	—	✅	—

---

🚀 Key Modules

🔐 Authentication & Role Management

• Secure JWT-based authentication
• Three distinct roles: Faculty, Student, Personal User
• Role-based UI and API access control

---

👩‍🏫 Faculty Module

• Create classroom sections with unique access codes
• Upload reference assignments per section
• View and review all student submissions
• Detect:
  • Student-to-reference plagiarism
  • Student-to-student cross-comparison
• Add remarks, grades, and resubmission requests

---

👨‍🎓 Student Module

• Join sections via access code
• Upload assignments (.pdf / .docx)
• View plagiarism percentage and AI detection results
• Track submission attempts, deadlines, and faculty feedback

---

📄 Personal Document Checker

Upload any academic document and get a detailed integrity report:

• Domain-based analysis: Computer Science · Science · Commerce · General
• Fingerprint-based plagiarism detection against a public domain corpus
• AI content detection using a transformer model
• Severity classification: Low · Medium · High
• Highlighted matched phrases in results
• Downloadable PDF report — professional and shareable

---

🧠 Detection Methodology

✔ Fingerprint-Based Plagiarism Detection

Text Normalization → Word Shingling → SHA-256 Hashing → Corpus Comparison → Similarity Score

1. Text is cleaned and broken into overlapping word shingles
2. Each shingle is hashed using SHA-256
3. Hashes are compared against a domain-specific public corpus
4. Similarity is derived from the fingerprint overlap ratio

📊 Severity Thresholds

Similarity Score	Severity	Interpretation
≤ 15%	🟢 Low	Common academic phrasing
16–40%	🟡 Medium	Partial content resemblance
> 40%	🔴 High	Strong overlap — likely plagiarism

---

🤖 AI Content Detection

• Model: roberta-base-openai-detector (HuggingFace Transformers)
• Output: AI Generated / Human Written + Confidence Score
• Integrated directly into the results table for both personal and classroom uploads

---

📑 PDF Report Generation

Auto-generated reports include:

• Document filename & upload timestamp
• Plagiarism percentage + severity badge
• AI detection result + confidence score
• Highlighted matched phrases
• Downloadable directly from the Results page

---

🧰 Tech Stack

Frontend

• React.js — Component-based SPA
• Tailwind CSS — Utility-first styling
• Fetch API — REST communication
• Toast Notifications — User feedback

Backend

• FastAPI — High-performance Python API
• SQLAlchemy — ORM for database interaction
• PostgreSQL — Relational data storage
• JWT — Stateless authentication

AI / ML

• Scikit-learn — TF-IDF & cosine similarity
• HuggingFace Transformers — AI detection model
• SHA-256 Fingerprint Hashing — Custom plagiarism engine

Cloud & Utilities

• AWS S3 — Secure file storage
• ReportLab — PDF generation

---

📁 Project Structure

EduShield/
├── backend/
│   ├── main.py           # API routes & app entry point
│   ├── models.py         # SQLAlchemy database models
│   ├── schemas.py        # Pydantic request/response schemas
│   └── database.py       # DB connection & session config
├── frontend/
│   ├── src/
│   │   ├── pages/        # Route-level page components
│   │   ├── components/   # Reusable UI components
│   │   └── App.jsx       # Root component & routing
│   └── public/
├── uploads/              # Temporary file storage
└── README.md

---

⚙️ Local Setup

Prerequisites

• Python 3.9+
• Node.js 18+
• PostgreSQL
• AWS S3 bucket (for file storage)

Backend

cd backend
python -m venv venv
venv\Scripts\activate          # Windows
# source venv/bin/activate     # macOS/Linux

pip install -r requirements.txt
uvicorn main:app --reload

API available at http://localhost:8000
Interactive docs at http://localhost:8000/docs

Frontend

cd frontend
npm install
npm run dev

App available at http://localhost:3000

---

🔒 Access & Licensing

This repository is private. Access is granted only to collaborators explicitly added by the owner.

---

📈 Roadmap

• Online corpus integration for broader plagiarism coverage
• Semantic plagiarism detection (beyond exact fingerprints)
• Admin analytics dashboard
• Multi-language document support
• Cloud deployment (AWS / Azure)
• Email notifications for deadlines and feedback

---

👩‍💻 Author

Samiksha Patil
BSc Information Technology — Final Year Project
🛡️ EduShield — Integrity, Verified.