SpamGuard-NLP

A multilingual SMS spam detection system using NLP and machine learning with real-time API inference.

---

live demo : https://spam-guard-nlp.vercel.app

---

Problem Statement

In today’s digital communication landscape, SMS and messaging platforms are increasingly targeted by spam, phishing attempts, and fraudulent content. Traditional spam detection systems are often limited to English-only datasets and struggle to handle multilingual and code-mixed messages, which are common in real-world scenarios—especially in regions like India.

Spam messages not only clutter user inboxes but also pose serious risks such as:

• Financial fraud
• Phishing attacks
• Data privacy breaches

---

Objective

The goal of this project is to develop a multilingual SMS spam detection system that:

• Classifies messages as Spam or Ham
• Handles multiple languages via translation and normalization
• Provides real-time predictions through an API
• Enhances interpretability with confidence scores and contextual insights

---

Solution Overview

SpamGuard-NLP is an end-to-end machine learning application with a Flask-based API and web interface that integrates:

• A complete data processing and training pipeline
• A modular prediction service layer
• A Flask-based REST API for real-time predictions

---

Pipeline Flow Diagram

---

Features

• Multilingual support via translation (English, Spanish, Arabic, etc.)
• Integrated translation for multilingual input handling and normalization
• NLP preprocessing
• TF-IDF feature extraction
• Multinomial Naive Bayes classifier
• Real-time prediction via Flask API
• Provides confidence scores and additional insights for predictions
• Modular and extensible architecture
• Structured logging for monitoring and debugging

---

Tech Stack

Languages & Libraries

• Python
• scikit-learn
• NLTK
• pandas, numpy

Backend

• Flask

ML Techniques

• TF-IDF Vectorization
• Multinomial Naive Bayes

---

Project Structure

SpamGuard-NLP/
│
├── assets/
│   ├── logs.png
│   ├── output.gif
│   ├── pipeline.png
│
├── src/
│   ├── __init__.py
│   ├── config.py
│   ├── data_loader.py
│   ├── evaluate.py
│   ├── logging.py
│   ├── model.py
│   ├── prediction_service.py
│   ├── preprocessing.py
│   ├── training.py
│   ├── translation.py
│   ├── utils.py
│   ├── vectorizer.py
│
├── tests/
│   └── test_prediction_service.py
│
├── static/
│   ├── app.js
│   └── style.css
│
├── templates/
│   └── index.html
│
├── artifacts/          # Saved model (generated)
├── data/               # Dataset (if included)
├── logs/               # Logs (ignored in git)
│
├── app.py              # Flask API
├── main.py             # Training pipeline
├── requirements.txt
├── .gitignore
├── LICENSE
└── README.md

---

How It Works

1. User inputs a message and selects a language
2. Message is translated (if required)
3. Text is cleaned and preprocessed
4. TF-IDF converts text into numerical features
5. Model predicts Spam or Ham
6. API returns:
   • Label
   • Confidence score
   • Normalized text
   • Translated text
   • Tips

---

API Endpoint

POST /predict

Request:

{
  "message": "You have won a prize!",
  "language": "en"
}

Response:

{
  "ok": true,
  "result": {
    "label": "Spam",
    "confidence": 98.2,
    "tips": "...",
    "translated_text": "...",
    "normalized_text": "...",
    "language": "en"
  }
}

Logging

Structured logging is implemented to track:

• API requests
• Predictions
• Errors

Logs are stored in:

logs/app.log

Run locally

pip install -r requirements.txt
python -m nltk.downloader stopwords
python app.py

Then open http://127.0.0.1:5000.

Train the model

python main.py

Training creates artifacts/model.pkl and artifacts/vectorizer.pkl.

Future Improvements

• Deep learning models (BERT, LSTM)
• Cloud deployment
• Improved multilingual support