Read and explore this rich interactive of 20 *million* research articles from PubMed, a project we're releasing today with Rita González Márquez @ritagonmar.bsky.social and Dmitry Kobak. It's a *beautiful* embedding structure, a fascinating, complete corpus. Some highlights (thread)

@hippopedoid 1. This presents some really interesting ways to take an *entire* digital library and make it searchable and browsable. Showing ten search results at a time in a browser is *not* the right way to see what's in a corpus: here you can start from global structure first and zoom in.

2. This is a static webpage! The search here runs entirely in-browser so it's lightning-fast: by sending @ApacheArrow buffers to the browser, and rendering and processing with deepscatter, any laptop can fetch, show, and search millions of titles locally.

3. By making the points the individual articles, they can really dance. One of my favorite nuggets @ritagonmar uncovered in her analysis was how abstracts *precisely* hit the word limits of 200, 250, etc; in browser you can identify the particular works!

The section towards the end about gender differentials--here, since it's static I can link it for you! offers a great canvas to explore which areas of medicine and biological research remain male-dominated even as the field becomes more equitable.

This was super fun for me to *write* because I got to use my set of svelte components implementing pandoc's document definition--that meant we could collaboratively write the story and plots in a single markdown file and recompile

The web site here is the online companion to an article released on biorxiv by @ritagonmar, @CellTypist, @hippopedoid, and others: Dmitry gives a description here.

@elotroalex points up the gender link doesn't work. That's Twitter's fault; it added some extra characters from the Tweet to the end of the URL. (Well, or friends don't let friends overuse em-dashes.) Correct link:

@ritagonmar @hippopedoid Wow this is super cool!

@ritagonmar @hippopedoid Nice work!

@ritagonmar @hippopedoid Wow, really amazing thank you for sharing this!