Design a friend recommendation system for Facebook.

Clarification question:

Am I designing a brand new friend recommendation system or improvising the already existing one? The Former.

User Groups:

Since the friend recommendation system is a must-have thing for FB and is common for everyone in the social media platform, there is no need to think differently for different user segments.

 Let A and B be Facebook friends, and we are trying to build a recommendation system to decide whether to suggest C as an option @ the People you may know the list.

1. If A and B have a mutual friend C, there is a high chance that person C might be a friend of A. 
2. If A and C have gone to the same school at the same time period, i.e, a difference in time of about +2 or -2 years, person C might be a good option to suggest. This is because a person who passed out of school in 2015 might not know someone who entered the same school @ 2015 for his/her sixth standard
3. If A and C have gone/goes to the same workplace at the same location, we have a good chance of both of them being friends.
4. If A and C have reacted to the same posts for a long time and thus showing the same interest can be paired out as a friend. However, this has the least priority as they might not know each other.
5. If A and C have some mutual private groups, they can be paired as friends and be suggested to others. For example, some alumni group.
6. If A has searched and viewed a profile C, then they could be shown as the fact that they have searched and viewed shows they are interested in and know each other somehow.
7. If both A and C are tagged in a photo of a post of some friend.