SpamShield – AI-Powered SMS Spam Detection

Project Overview

SpamShield is a full-stack AI-powered web application that detects whether an SMS message is Spam or Ham (Safe) using Natural Language Processing and Machine Learning.

The project began as an NLP classification model and was later transformed into a production-ready full-stack web application with a modern responsive user interface.

Users can paste suspicious SMS messages and instantly receive spam detection predictions through a live deployed web application.

---

Live Demo

🚀 https://sms-spam-detector-kzfg.onrender.com

---

Features

• AI-powered SMS spam detection
• Real-time message classification
• NLP text preprocessing pipeline
• TF-IDF vectorization
• Multinomial Naive Bayes classification
• REST API backend using Django REST Framework
• Modern responsive frontend built with React
• Beautiful UI designed using Tailwind CSS + Stitch-inspired design
• Full-stack deployment on Render
• Production-ready frontend/backend integration

---

Tech Stack

Frontend

• React.js
• Axios
• Tailwind CSS
• Lucide React Icons
• Vite

Backend

• Python
• Django
• Django REST Framework
• django-cors-headers

Machine Learning / NLP

• Pandas
• NumPy
• NLTK
• Scikit-learn
• TF-IDF Vectorizer
• Multinomial Naive Bayes

Deployment

• Render

---

Dataset

SMS Spam Collection Dataset

Dataset contains:

• 5,572 SMS messages
• Real-world spam/ham labeled messages
• Binary classification dataset

---

NLP Pipeline

SMS Message Input
→ Lowercase Conversion
→ Remove Numbers
→ Remove Punctuation
→ Stopword Removal
→ TF-IDF Vectorization
→ Multinomial Naive Bayes Prediction
→ Spam / Ham Output

---

Model Performance

Accuracy: 97.13%

Model used:

Multinomial Naive Bayes

---

UI Preview

Web Application Interface

---

Results

---

Example Predictions

Spam

Congratulations! You won a free iPhone. Click now!

Result:

🚨 Spam

Ham

Hey, where are you? Call me when you're free.

Result:

✅ Ham

---

Project Structure

spamshield/
│
├── backend/
│   ├── config/
│   ├── predictor/
│   ├── model.pkl
│   ├── vectorizer.pkl
│   ├── requirements.txt
│
├── frontend/
│   ├── src/
│   ├── public/
│   ├── package.json
│
├── spam_classifier.py
├── README.md

---

Run Locally

Backend Setup

cd backend
pip install -r requirements.txt
python manage.py runserver

Backend runs at:

http://127.0.0.1:8000

---

Frontend Setup

cd frontend
npm install
npm run dev

Frontend runs at:

http://localhost:5173

---

Production Deployment

Frontend + Backend deployed on Render

Deployment includes:

• React frontend build using Vite
• Django backend hosting
• Static asset serving
• ML model pickle loading
• API integration
• Production build automation

---

Future Improvements

• Email spam detection
• URL phishing detection
• User authentication
• Spam history dashboard
• Dark mode UI
• Mobile-first PWA version
• Confidence score display

---

Author

Siva Priyanka

GitHub: https://github.com/sivapriyanka-dev

Passionate about Machine Learning, NLP, full-stack development, and building real-world AI applications 🚀