Some background: https://www.infodocket.com/2020/05/07/machine-learning-the-l...

The dataset itself: https://news-navigator.labs.loc.gov/

You should also try Trove's newspaper archives. See https://trove.nla.gov.au/newspaper/

And the Library of Congress' own newspaper archives: https://chroniclingamerica.loc.gov/

There's some fascinating stuff in there, like this from the Bamberg Herald in November 1918: https://chroniclingamerica.loc.gov/lccn/sn86063790/1918-11-2...

Wonder what the print run was on that day, lol

That's the dataset used in the paper.

Recenty my browser's home page is a real-time stream of randomly clipped newspaper articles going back to the 1600s: https://newspaperarchive.com/clippings/

It's fun to see what people are digging into the newspaper archives for, and to speculate how it may relate to present or future events.

Note: paid subscribers see more (which I am for family history purposes).