Spamizard

Spam meets Wizardry

Spamizard is a full-stack machine learning application designed to classify emails as Spam or Ham (Not Spam) and automatically generate intelligent replies.

The Model

• Algorithm: Utilizes a Multinomial Naive Bayes classifier, which is highly effective for text classification tasks based on word frequencies.
• Feature Extraction: Uses TF-IDF (Term Frequency-Inverse Document Frequency) vectorization to map textual email data into numerical features, heavily weighting unique words while discounting common stop-words.
• Training: Trained on a robust dataset of over 5,000 labeled emails, achieving over 96% accuracy on unseen validation data.

The Web Interface

The model is deployed via a Flask backend with a responsive, modern web interface. Users can paste in raw email content, and the server runs real-time inference to determine if the email is a threat or a genuine message. If it’s genuine, the system provides an AI-assisted quick reply template.