Matched to You feed sorting orders the papers in a feed based on your previous interactions. This only changes the order and does not change the papers within the feed. This can save time when looking at a feed with many papers or when looking at the week or month view.
The Matched to You approach is based on papers you click within a specific feed. Once you click on 20-30 papers, we use those papers to create a representation of what you would like to read in this feed. Then when we get new papers for the feed, we compare each paper to this representation and put the papers most similar at the top. We periodically recalculate the personalized representation in order to incorporate new papers.
At Meta, we are cognizant of both the benefits and risks of personalization. Personalization can save time and help you find relevant information faster. On the other hand, there can be a bubble effect where content in a newsfeed becomes more similar over time. As a scientific community, we must be aware of this risk. To combat the bubble effect, Meta has an option to return to Default ranking. This can be useful if you would like to see papers that are not similar to your previous interactions. It can also be useful to re-train the personalized representation if you believe it has gotten off track. We update the representation of papers you like to see in this feed based on clicks from both the Default and Matched to You views, reducing the chance of ending up in an information bubble.
If you are not able to see Matched to You sorting option for a feed, this is because we have not seen enough clicks on papers in the feed to generate the personalized representation of what you would like to see. In addition, some ad blockers stop this information from coming through, so please disable your ad blocker.
To learn more about the Matched to You sorting algorithm, please read our blog post.