Analysis:

Let's look at all books in library at a glance. Basically, I ran each of 65k book titles in library through sota jina-v3 multilingual sentence transformer, then PCA to reduce those dimensions, then UMAP to further reduce to 2 dimensional plot, then hdbscan to cluster them with intuitive labels

more to come...