This is a map of the 1000 most mentioned books on Hacker News. You can also see it as a list.

I scraped 20k posts related to books and reading from Algolia. I used GPT-4o to parse the posts and extract references to books, along with the opinions of the posters.

These were collated into descriptions, which were then embedded using OpenAI’s model. The embeddings were projected into 2D using PaCMAP and clustered with HDBSCAN. Cluster labels were generated with GPT as well.

The clusters were packed as circles to get an aesthetically pleasing result. The corresponding covers were aligned on a staggered grid, stitched together, and visualised using OpenLayers.