Now that we've seen how this process works in general, let's go into it a little more step by step
Posted: Sun Jan 19, 2025 8:18 am
Inventory
As we have mentioned, in this first phase all the publications that could reach the user's profile are collected, of which there are thousands for each of the 2 billion users. Facebook exemplifies this process through a user called Juan.
«This eligible inventory includes any posts shared with John by a friend, group, or Page he's connected to that have been made since his last login and haven't been deleted. But how should we handle posts created before John's last login that he hasn't yet seen?
To ensure that unseen posts are reconsidered, new posts that were india business email database classified as eligible for John (but that he didn’t see) in his previous sessions are added to the eligible inventory for this session. We also apply logic so that any posts that John has already seen and that have sparked an interesting conversation among his friends are also added to the eligible inventory.
The system then has to score each post based on a variety of factors, such as the type of post, similarity to other items, and how closely the post matches what the user tends to engage with. To calculate this for over 1,000 posts, for each of billions of users, all in real time, Facebook runs these models for all candidate stories in parallel on multiple machines, called predictors.
As we have mentioned, in this first phase all the publications that could reach the user's profile are collected, of which there are thousands for each of the 2 billion users. Facebook exemplifies this process through a user called Juan.
«This eligible inventory includes any posts shared with John by a friend, group, or Page he's connected to that have been made since his last login and haven't been deleted. But how should we handle posts created before John's last login that he hasn't yet seen?
To ensure that unseen posts are reconsidered, new posts that were india business email database classified as eligible for John (but that he didn’t see) in his previous sessions are added to the eligible inventory for this session. We also apply logic so that any posts that John has already seen and that have sparked an interesting conversation among his friends are also added to the eligible inventory.
The system then has to score each post based on a variety of factors, such as the type of post, similarity to other items, and how closely the post matches what the user tends to engage with. To calculate this for over 1,000 posts, for each of billions of users, all in real time, Facebook runs these models for all candidate stories in parallel on multiple machines, called predictors.