Using CLIP, a multi-modal embedding model, I matched images to stanzas and created an engagement-based recommendation system. The algorithm averages your last ten viewed stanzas, weighted by viewing time, to find the next most similar content. It's simple, but (I think!) effective.

This project aims to demonstrate how AI and the humanities can complement each other, creating something meaningful when thoughtfully combined—rather than being at odds, as they’re often portrayed (or even often are).

Link to the repo: https://github.com/jswest/leaves-of-grasstok