Facebook Friend Suggestion Algorithm

Facebook?is a social networking service that allows you to interact and share information with your family and friends over the?internet. Facebook?is a social networking service that allows you to interact and share information with your family and friends over the?internet. Originally designed for college students, Facebook was created in 2004 by Mark Zuckerberg while he was enrolled at Harvard University. Today, Facebook is the world’s largest social network, with?more than?1 billion users?worldwide.

What makes Facebook unique is the ability to?connect and share?with the people you care about at the same time. Facebook allows you to?send?messages?and?post status updates?to keep in touch with friends and family. You can also share different types of content, like?photos?and?links. But what’s more interesting is the people suggestion feature that everyone sees on their Facebook where most common people are recommended for mutual to add on your friend list. So how does Facebook do this? Let's explore it.

How Does Facebook’s Suggested Friends Work?

Your friend suggestions are generated when one of your friends select you as someone who knows someone else on Facebook. If you add your suggested friends as friends, a normal friend request will be sent. If you do not, no one will be notified that you ignored a suggestion. The friendship suggestion is based on the Facebook algorithm, which considers a variety of issues, including previous connections, previous activities, and profile information. Facebook uses?Collaborative?& Content-based filtering?to recommend people you might know, display ads based on your posts, jobs you might like, groups you might want to follow, or companies you might be interested in. Facebook also uses?Apache Giraph to analyze the social graph formed by users and their connections. Facebook has been working on how to improve its algorithm to surface the best content to the people who are most likely to engage with it, which should lead to fewer interruptions for users. The Facebook algorithm ensures that all Facebook users get the most relevant updates, news, and information they are interested in. It is by no means an easy algorithm to crack, but some ranking factors are well known : inventory , signals , predictions and relevancy score.

Some of the ML techniques that Facebook uses are:

Collaborative & Content-Based filtering

Collaborative Filtering?is a?Machine Learning?technique for detecting connections between data sets. This method is often used in?recommender systems?to find similarities between user data and items. Content-based filtering?is a form of a?recommender system?that tries to predict what a user might like based on their?past behavior.?Content-based filtering generates suggestions by matching?keywords?and?attributes?related to database items to a user profile.

In?collaborative?based on the user's?previous?activity, the groups they follow on Facebook, kind of photos they are tagged in are the some of the factors that the?collaborative?algorithm?will consider and?suggest?a friend accordingly.

In?content-based?users’?predefined?data such as age, gender, location, and school/work based on this information the?content-based algorithm?will?suggest?a friend.

K-Mean Clustering

K-means?clustering is a widely used?clustering?method. In general, practitioners begin by learning about the dataset’s architecture. The K-means clustering algorithm divides data points into?distinct,?non-overlapping?groups.

For Facebook friend recommendation a user will get a?recommendation?from another user who belongs to a?similar?group. The?similarity?can be based on multiple things such as their interest, followings, location, background, etc considering all these factors a?cluster?can be formed.

Deep Text

While photographs may appear to be the most significant aspect of Facebook, writing is just as crucial. On Facebook, there is a lot of?writing. Facebook utilizes?DeepText, a?deep?learning-based text?engine?that can comprehend thousands of posts in a second in more than 20 languages with as much accuracy as you can!

We can recommend based on the?comments?of the users over a particular topic.?Deep?learning?can analyze these comments and give a?recommendation?based on their response. For example, if two users commented something?positive?on a topic or a post they both can be?recommended?to each other, not just positive even if two users gave a?negative?comment can be?recommended?to each other.

. . . .

The Facebook algorithm determines which posts people see every time they check their Facebook feed, and in what order those posts show up. The algorithm isn’t static. Meta has a whole team of people working on artificial intelligence and machine learning. Part of their work is to improve the algorithms that connect Facebook users with the most valuable content for them. It also determines the friend suggestions based on various factors. Some of the are:

• Friends you add:?once you add a person, you have given the Facebook a clue who you are interested in adding as a friend. So, Facebook will start looking for a similar profile, and people in common. The most probably, the suggestion would be changed to that’s users’ friends.
• Friends of friends: one of the most common ways that Facebook suggestion is mutual friends. You might face this and you see that sometimes you might have more than 100 mutual friends with someone.
• Bio: the way you fill-up the bio affect who is going to be your next friends. Based on the information of your schools, university, places you lived, and family members Facebook would find and suggest those who are in the same category.
• Likes and comments: the way you like posts affect the Facebook friend’s suggestion. For example, if you like a page related to the automobile industry, therefore people who might have the same interest as you would come into the friend suggestion list.
• The profile you visit: if you frequently visit a profile on Facebook, this would send an alarm to Facebook that you are interested in having that person as a friend. So, they would be added to the suggestion list.
• Facebook search bar: every term you search on Facebook search bar can be counted as a sign to show your needs. Facebook will get this soon and send you the suggestion that might help you with your needs. Other people might also search your profile so even that might show up on our suggested friends list.
• Google search: this is an estimated factor of Facebook suggestion because I have experience of this. Once I was looking for my favorite program and university on google, I noticed that the type of profile that Facebook shows to me is changing. I realized that those users were studied at the searched universities by me, or they have studied what I have explored.
• Contacts on the phone:?The people who already have your contact number saved in their Phonebook. Therefore, any person who has your contact number in their contacts will be visible in the “People you may know” section on your Facebook account as well as on Messenger.

Conclusion:

Facebook is constantly trying to acquire more information on its users to build a more accurate network graph, which will in turn lead way to better services like friend suggestions, event suggestions, and maybe even what shows up in your top news feed. Facebook is working on rolling out updates to its algorithm to promote more engaging content.

References:

1. https://www.socialchamp.io/blog/facebook-algorithm/
2. https://graziadaily.co.uk/life/real-life/facebook-suggested-friends-work/
3. https://guidesmania.com/does-facebook-suggest-friends/