Sumeet29/email-classification-system

Email Classification System

This is an intelligent email classification system that can:

1. 1.Classify emails into different categories
2. 2.Mask Personal Identifiable Information (PII)
3. 3.Provide both masked and demasked versions of the email

Features

• —Email Classification: Automatically categorizes emails based on their content
• —PII Masking: Detects and masks sensitive information like:
• —Names
• —Email addresses
• —Phone numbers
• —Credit card numbers
• —CVV numbers
• —Dates of birth
• —Aadhar numbers
• —Interactive Interface: Easy-to-use web interface for testing the system

How to Use

1. 1.Enter your email text in the input box
2. 2.Click "Submit"
3. 3.View the results showing:
4. 4.Original email
5. 5.Category classification
6. 6.Masked version (with PII hidden)
7. 7.Demasked version
8. 8.List of detected PII entities

Example

Try these example inputs:

Hello, my name is John Doe and my email is johndoe@example.com. I'm having trouble with my billing.

Hi, I need to reset my password. My phone number is 123-456-7890.

Please update my credit card ending in 1234. My CVV is 123.

Technical Details

• —Built with Python and Gradio
• —Uses machine learning for classification
• —Implements regex-based PII detection
• —Handles multiple types of sensitive information

Model Information

The system uses a trained classifier model (email_classifier.pkl) that categorizes emails based on their content. The model is loaded from the model directory.

Privacy

All processing is done locally in your browser. No data is stored or transmitted to external servers.

License

MIT License

Installation

1. 1.Clone the repository:

git clone <repository-url>
cd email-classification-system

1. 1.Create a virtual environment and activate it:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

1. 1.Install dependencies:

pip install -r requirements.txt

Usage

Training a Model

To train a new model:

python app.py --mode train --data_path data/emails.csv --model_type traditional --classifier_type naive_bayes

Evaluating a Model

To evaluate an existing model:

python app.py --mode evaluate --data_path data/emails.csv --model_path model/email_classifier.pkl

Running the API

To start the API server:

python app.py --mode api --model_path model/email_classifier.pkl

The API will be available at http://localhost:8000

Testing the API

To test the API:

python test.py

API Endpoints

• —POST /classify_email: Classify an email
• —GET /health: Check API health status

Data Format

The input CSV file should have the following columns:

• —email: The email text
• —type: The category of the email (for training)

Model Types

1. 1.Traditional ML Models:
2. 2.Naive Bayes
3. 3.Random Forest
4. 4.SVM

1. 1.Transformer Models:
2. 2.DistilBERT (default)
3. 3.Other Hugging Face models can be used

PII Masking

The system masks the following types of PII:

• —Full names
• —Email addresses
• —Phone numbers
• —Dates of birth
• —Aadhar numbers
• —Credit/Debit card numbers
• —CVV numbers
• —Expiry dates