Facebook’s Trending Algorithm

Ever wonder how, exactly, Facebook’s trending algorithm works: how it chooses its “Trending” news stories? Those of you who use Twitter and frequent its “Trends” section will find that the two social media giants share very few, if any, trending topics, or even methods to their “Trending” madness.

Facebook’s Trending algorithm is different from Twitter’s. And that’s okay. They both have their own strengths. Facebook’s Trending topics are basically a collection of today’s Internet news headlines, whereas Twitter’s Trends are what is being talked about right now. Where Facebook lacks in immediate relevance, it makes up for in presentation of its Trending stories. Twitter’s trends are immediately relevant, but they are just a one- or two-word link to any and all Tweets that contain that hashtag, name, or phrase.

Facebook’s Trending Algorithm: Something out of a Sci-Fi Novel.

What would be your best guess as to how Facebook’s Trending algorithm works? How does it come up with its Trending stories? An Editing Room full of Millenials on iMacs sifting through all the WaPo, Huff Post, and BuzzFeed articles on the internet? Or a similar room full of Millenials looking through all its users’ posts on iMacs? It has to involve Millenials and iMacs, right?

Wrong. Well, at least about the Millenials and the Internet articles. And the iMacs.

Like Twitter, Facebook uses its own algorithm to identify trending topics, albeit a much more sophisticated and impressive one. Facebook automatically analyzes every post, story, and status that its users, groups, and Pages post to the social media platform. But, without hashtags, how does Facebook figure out the topic of each post?
Facebook’s Trending algorithm uses a system called Natural Language Processing, a combination of “computer science, artificial intelligence, and computational linguistics concerned with the interactions between computers and human (natural) languages,” according to Wikipedia. Essentially, Facebook translates every post, from every user, in every language, to a computer language it understands, so that it can identify the topics and substances of all the chatter amongst its users, and compare them to current and recent events. Unlike Twitter, which more or less just posts the most-mentioned hashtags or phrases to its “Trends,” Facebook Trending topics must actually be tied to an event that took place, or is currently happening.

Why is a language algorithm necessary? With Twitter, hashtags are essentially designated catch phrases that anyone from companies, ad campaigns, events, to TV shows can use to loosely group Tweets in to that campaign’s discussion. Hashtags also arise and populate in real time among user discussion. Hashtags also serve to identify the topic being discussed. Facebook does this on a grander, more complicated scale, and does the whole process in reverse. It takes all Facebook posts, catalogs all of them, uses language software to identify the topic, and sees if that topic applies to any events that are currently or recently occurring. If enough posts come out identifying a new topic, it is very likely that a current event is being talked about.

The Factors Involved in Choosing Your “Trending” List

How does a trending topic make it to your “Trending” list? Facebook is pretty vague about the user and topic factors it takes into consideration; Facebook says it takes into account factors such as “engagement [times you’ve liked or commented on a certain topic], timeliness [time relevance], Pages you’ve liked and your location. (See “How does Facebook determine what topics are trending?”)”  So, if you like your sister’s custom shoe startup Page, are you more likely to see news stories about Nike? No one really knows for sure. One of the strongest factors you can always count on, however, is location.

So a new trending topic is being discussed and made its way into Facebook’s Trending algorithm. Does it make it to your Trending section? If the event or news story is local to you, and matches your interests (if it’s discussed enough, widely enough, these things seem to be less of factors), it may be identified as a “Trending” topic to you, the possibly interested user. The more discussed the topic and the more widely it applies, the wider the area it may apply to. If a topic is being talked about enough all over a country, it will probably be a national “Trending” topic.

Facebook says you can also customize your Trending section the same you do ads and stories you are interested in: if a topic comes up that you have no interest in, hover over it and click on the “x”, and Facebook will “learn” that such topics shouldn’t come up again.

Human-Written Descriptions, Computer-Generated “Trending” Story Feed

It is commonly speculated that people at Facebook choose Trending topics, as each obviously has a human descriptor, but Facebook assures us this is not the case. There is a team that writes a short description of a topic, and identifies the event taking place that is the trending topic (remember, each “Trending” topic must be tied to an event), but that is all the human interference that takes place.

When a trending topic is clicked on, a page, not unlike a news feed, is automatically compiled with a longer description, a “Top Posts” section which are usually news articles, if any have yet been written, and a “Public Posts” section, which is just that, posts by the average Facebook user that mention the trending topic.

Next Up: Twitter’s “Trends” algorithm, and Facebook’s “Trending” vs. Twitter’s “Trends”