Algorithm:
The algorithm embeds each piece of content as a 256 dimension vector, each essay gets one of these. When a user interacts with essays in a good way, the user becomes more similar to the essay and the essays interaction vector becomes more similar to the user.
Best content is found by finding the most similar essay and essay interaction vectors to the user. This way finding new content (due to interaction vectors in the essay) becomes possible and it is likely to be great.
Algorithm: The algorithm embeds each piece of content as a 256 dimension vector, each essay gets one of these. When a user interacts with essays in a good way, the user becomes more similar to the essay and the essays interaction vector becomes more similar to the user. Best content is found by finding the most similar essay and essay interaction vectors to the user. This way finding new content (due to interaction vectors in the essay) becomes possible and it is likely to be great.