talib-ai-ml/Email-Classification

Email Classification for Support Team

Overview

This project implements an email classification system for support teams, featuring:

1. 1.PII Masking: Detects and masks sensitive information (e.g., names, emails, credit card numbers) using regex.
2. 2.Email Classification: Uses a RandomForestClassifier trained on TF-IDF vectorized email text to categorize emails into 4 classes: Incident, Request, Problem, Change.
3. 3.API Deployment: FastAPI endpoint that accepts an email and returns the masked text, PII entities, and predicted category.

Setup

Install dependencies:

pip install -r requirements.txt

Run the API locally:

uvicorn api:app --reload

Access the API at: http://127.0.0.1:8000/docs for interactive testing.

Deploy on Hugging Face Spaces

• —Add api.py, models.py, utils.py, and model files (*.pkl) to the Space.
• —Configure the Space to use FastAPI (an example Dockerfile is provided in the repo).

API Usage

Send a POST request to /classify_email with a JSON payload:

{
  "email_body": "My name is John Doe. My email is john@example.com. I need help with my account."
}

Response Format:

{
  "input_email_body": "Original email text",
  "list_of_masked_entities": [
    {"position": [start, end], "classification": "full_name", "entity": "John Doe"},
    {"position": [start, end], "classification": "email", "entity": "john@example.com"}
  ],
  "masked_email": "My name is [full_name]. My email is [email]...",
  "category_of_the_email": "Request"
}

Files

• —api.py: FastAPI endpoint.
• —models.py: Loads the trained model and vectorizer.
• —utils.py: PII masking/unmasking logic.
• —email_classifier_model.pkl, vectorizer.pkl: Pretrained model files.

Run using:

python -m uvicorn api:app --reload