jev-spam-eval

What it does

This notebook compares zero-shot spam judgments from Jev with conventional TF-IDF classifiers.

How you can use it

Your developer can adapt this experiment to sort incoming messages into clear categories like spam or legitimate mail. The app extracts details like the sender, message text, and hidden web addresses. It sends this evidence to TypeSafe, an outside AI service, to choose the right category.

A developer can review the included Python files to see exactly how the questions are worded. They will need an access key that connects the app to TypeSafe. The AI service only reads the supplied text. It does not click the links or open the file attachments.