NewsChain uses 10,000 pages of scraped BuzzFeed headlines as input for a Markov chain to generate new ones. It then searches those headlines on Google Images and downloads the first ten Creative Commons-licensed results, then performs reverse image searches on those to get captions.

"Articles" are saved to a JSON file with an MD5 hash to identify them and tie them to their images. GET requests to /refresh/ add another article. Article thumbnails are available on the main page, linking to

Built With

Share this project: