An adaptation of PageRank – EdgeRank
Article source: http://edgerank.net/
https://marketingland.com/edgerank-is-dead-facebooks-news-feed-algorithm-now-has-close-to-100k-weight-factors-55908
An interesting application and variation of the PageRank algorithm implemented by Google in ranking search results is the EdgeRank algorithm Facebook used to decide what is displayed on each user’s newsfeed. Both algorithms accomplish the same principle goal: the more closely connected and relevant something is, the higher ranked it should be when displayed to the user.
Facebook’s EdgeRank was based on three main factors: affinity score, edge weight, and time decay. Affinity score is similar to how the PageRank algorithm works. It measures how connected a person is to an Edge (an Edge is anything that can potentially be displayed on your newsfeed). In the PageRank algorithm if a page is connected/linking to another page, then it passes it’s PageRank value onto that page. Similarly, if a user has many interactions (or connections) with an Edge then that Edge will have a greater value conferred onto it. However, a difference for Facebook is how connections are defined versus in Google’s search rankings. A connection or interaction on Facebook can be anything like commenting, liking/reacting, clicking links, tagging, sharing, etc. The people you have the most interactions with on Facebook tend to be the people you are closest to in real life or are most interested in, so it lends itself well to ranking what goes onto your newsfeed.
Edge weight is the second factor EdgeRank takes into account. Edge weight is based on how much effort you take in your interactions with the Edge. For example, commenting takes considerably more effort than just clicking like on a Facebook status post. This tells Facebook that you are more interested in this type of content or this person’s content, so it will rank it higher in it’s EdgeRank algorithms.
The last factor EdgeRank takes into account is simply time decay. As posts get older, they are less relevant to its users so this factor keeps the posts that are shown on your newsfeed fresh and prevents old and no longer relevant posts from appearing on your newsfeed.
Facebook has since moved on from the EdgeRank algorithm as machine learning plays a larger and larger role in solving problems without a well-defined solution. Facebook now uses a nameless machine learning algorithm that takes into account over 100,000 different factors when deciding what to display on each user’s newsfeed. However, the influences of EdgeRank still remain as the three EdgeRank factors still play a significant factor in the new newsfeed ranking algorithm.