Facebook People You May Know AI system

UPDATED DEC 13, 2024
The content you see on your Facebook People You May Know is selected, ranked and delivered to you by an artificial intelligence (AI) system. Within one AI system, multiple machine learning models work together to deliver your experience. These models and their input signals are dynamic and they change frequently as the system learns and improves over time.

Overview of Facebook People You May Know

When you view and interact with Facebook, one of the underlying AI systems delivers personalized recommendations of people you may know. These suggestions help you discover people you might want to add as Facebook friends.
How Facebook People You May Know works
The AI system behind People You May Know automatically recommends the profiles of people you might want to connect with. Here’s how it works:
  1. Gather inventory
    First, the system gathers profiles of people that might interest you. These could include people who are friends of your existing friends on Facebook or people in groups you’re a member of.
  2. Rank profiles
    The system calculates a score for each profile based on how likely you are to engage with it. Then the system puts these profiles in order by their scores. Profiles of people that the system predicts you’re likely to interact with are shown to you first.
  3. Apply integrity filters
    Finally, the system applies filters to remove profiles that may go against our Community Standards.
How to customize what you see
Your experience on Facebook People You May Know is personalized based on your activity, and you have options to control or customize what you see. Below, we describe how to do this with different in-product features. Options shown here may not be available to everyone.
Hide People You May Know
If you want to see the People You May Know unit in your Feed less frequently, you can hide it. This control is in the settings menu of the People You May Know unit itself.
Choose who can send friend requests to you
To change who can add you as a friend, you can select “Everyone” or limit to “Friends of friends.” Choosing “Friends of friends” also limits your own profile’s visibility to only friends of friends in People You May Know suggestions.
Dismiss suggestion
You can dismiss People You May Know suggestions in your Feed by clicking “X” in the upper right-hand corner or “Remove” under a single suggestion.
Discover people in a non-personalized way
To discover people on Facebook in a way that’s not personalized specifically to you, use the search feature in the Feeds tab. Learn more.
How the AI delivers content to you
We want you to see content you enjoy and find interesting. To achieve this, the AI system has models that help it make predictions about content you'll find most relevant and valuable. These prediction models use underlying input signals to help select content you're most likely to engage with.
Below are some of the significant predictions–and input signals that inform them–that we use in this AI system.
How likely you are to send a friend request to a person being suggested to you, if you have more than 60 Facebook friends or you use Facebook daily
Signals influencing this prediction include:
  • How many times you’ve sent friend requests to people who have posted a similar number of times or have a similar number of profile and cover photos as the suggested person has
  • How many conversions (e.g., likes, comments, shares) made by a user on Facebook over the past 7 days, with more recent conversions weighted more heavily
  • How many friend suggestions you’ve seen overall in certain time periods
  • Whether or not your contact was uploaded by the person being suggested
  • The estimated strength of the connections between the suggested person and their current and former friends, as well as the number of friend requests that person has declined
How likely you are to send a friend request to a person being suggested to you and then send more friend requests with the same PYMK carousel
Signals influencing this prediction include:
  • How many days in the last 28 days that a user has taken some action on Facebook
  • What percentage of the time others view the profile of the person being suggested when that person is suggested
  • Friend counts from the suggested user
  • Friend counts from the viewer
  • What percentage of the time others view the profile of the person being suggested when that person is suggested
How likely it is that the person receiving your friend request will accept it, if you have fewer than 60 Facebook friends and you do not use Facebook daily
Signals influencing this prediction include:
  • How many pending friend requests the person being suggested to you has
  • The number of friends of the person being suggested that react to posts
  • How long it's been since the suggested person was active on Facebook
  • Sum the ratio of accept scores to a dominator that is calculated based on the number of edges. The sum is calculated over all PYMK edges suggestions on Facebook
  • How many pending friend requests the person being suggested to you has
How likely it is that the person receiving your friend request will accept it, if you are in United States (US)
Signals influencing this prediction include:
  • How many pending friend requests the person being suggested to you has
  • What percentage of the suggested user to incoming friend requests that were accepted
  • Average recency of friend connections made by a user on Facebook, with more recent connections weighted more heavily
  • How many friend requests received by the suggested user that were accepted
  • What percentage of friend requests received by the suggested person that were rejected
How likely you are to send a friend request to a person being suggested to you, if you have fewer than 60 Facebook friends and you do not use Facebook daily
Signals influencing this prediction include:
  • How many friend suggestions you’ve seen overall in certain time periods
  • How many friend suggestions you’ve seen overall in certain time periods
  • What percentage of the time others view the profile of the person being suggested when that person is suggested
  • The estimated strength of the connections between the suggested person and their current and former friends, as well as the number of friend requests that person has declined
  • How many friend requests you have send that are currently pending
How likely you are to send a friend request to a person being suggested to you, if you are in United States (US)
Signals influencing this prediction include:
  • How many conversions (e.g., likes, comments, shares) made by the person over the past 30 days, with more recent conversions weighted more heavily
  • What percentage of the time others view the profile of the person being suggested when that person is suggested
  • How many friend suggestions you’ve seen overall in certain time periods
  • How many friend connections the suggested user has initiated and accepted by others
  • What ratio of number of mutual friends to total number of friends between two users on Facebook
How likely it is that the person receiving your friend request will accept it, if you have more than 60 Facebook friends or you use Facebook daily
Signals influencing this prediction include:
  • How long it's been since the suggested person was active on Facebook
  • How many meaningful social interactions the person being suggested to you has had over the last 30 days
  • How many friend requests have been accepted by the suggested person over certain time windows
  • How many pending friend requests the person being suggested to you has
  • How many friend requests have been accepted by the suggested person over certain time windows