01 / CONTEXT
The problem to solve
Explore a basic text-classification workflow for distinguishing spam from legitimate email.
02 / ENGINEERING
From need to system
- Convert email text to count-based features with CountVectorizer.
- Train a Multinomial Naive Bayes classifier.
- Expose predictions through a small Tkinter interface.
- Keep the example explicitly framed as a historical learning prototype.
03 / DESIGN DECISION
A meaningful choice
Choose a simple, interpretable baseline to make the classical NLP pipeline easy to follow.
04 / EVIDENCE AND LIMITATIONS
What is actually documented
The public README documents the 2019 prototype and its vectorizer, classifier and desktop UI. It is legacy code with no current benchmark or maintained deployment.
05 / NEXT VALIDATIONS
How I would strengthen the proof
- Add a reproducible train/test split and evaluation metrics.
- Review privacy and dataset provenance before reuse.
- Update dependencies if the prototype is revived.
These are planned validation steps, not already measured results.