As soon as you log in, news feed is the first page that you see. Ever wondered how news feed shows you the relevant content? Do you think its some random event? If facebook had showed you all contents generated by your friends, well, you know what would have happened. Especially when it comes to brands, facebook ensures that specific brands are seen by the right people through news feeds. Facebook recently showed us some insight into the algorithm that credits directly to the ‘News Feed Optimization’, the edgerank algorithm.
There are two distinct parts of the facebook’s news feed. 1.’Top news’ and 2. ‘Most recent’ . The news feed that you get to see, is always, by default, the top news. According to facebook, top news is “based on an algorithm[that] uses factors such as how many friends are commenting on a Post to aggregate content that [a person] will find interesting. It displays stories based on their relevance rather than in chronological order.”
On the shallow level, there exists something called an object.
Facebook views all the inputs as “Objects”. And those “objects” may be, status updates, video links, pictures, etc.
When you have an ‘object’ in your news feed, whenever some other user interacts with this object, ( say he/she comments on a status update) they are creating what is known as an Edge.
The edge rank formula:
$\sum_{edge e}U_e W_e D_e$
The image below shows the News feed optimization expression.
It is the snapshot of the algorithm, posted by a researcher at MIT.
There are three main components related to an Edge: Affinity, weight and time.
Affinity: You will have more affinity towards some friends on facebook than others. The more often you like, tag, comment, view, and as well click on a friend’s object, the so-called affinity score between you and your friend increases. And the affinity is always a one way path. If your friend has liked or commented on your posts, then there is a high probability that your news feed has your friend’s posts.
For example, I like The Beatles and I frequently visit its pages and comment on its posts. So, my news feed has its contents because of the high affinity I have to it, on facebook.
Weight: This measure deals with the value of an object’s edges. More the number of edges an object has, more is its weight and so will be its Edge rank. An edge differs from the other based on whether its created due to commenting or liking or tagging that post.
It is obvious that more weight will be given to an edge if it’s a comment on the respective object compared to when it is liked.
Time: Your news feed is most likely to contain today’s news rather than yesterday’s. Hence facebook has this criteria add to its edge rank. Affinity and weight are not the only criteria that facebook looks into. Because time is the most valuable factor that might change the way your pages look. For example if Karnatka’s CM has been replaced on Wednesday and it is the top most news, everyone is commenting on, and let say it has received some 500 comments, it won’t be on your news feed after a week despite of having large number of comments. ‘Time decay score’ is the major factor here playing its role and the reason behind our news feed getting the latest news.
Now, is this really important?
If it wasn’t showing you relevant information, then the chances of you missing on something really important would be high. Or you might just become bored with all kinds of news, all the time, both relevant and irrelevant which might become one of the reasons for facebook’s breakdown!
When it comes to business and brands, a user’s affinity towards your brand is your key ingredient for success. If you want your brand’s contents on almost everyone’s news feed then you have to create some, that will have high edge rank based on all the three components.
A white paper has been written to explain Facebook's edge rank algorithm. You can refer to it for more information.
http://forms.buddymedia.com/whitepaper-form_facebooks-edgerank.html
Happy facebooking :-)
Shruti Ranade
'Well Begun' is half done !! What a topic !! :) Thrilled to know the internal working of the facebook :)Affinity well explained. It reminds me of the 'preferential attachment' factor that has a major contribution in the formation of scale free networks :) :) Haha..Time criteria explained nicely..:) very informative blog.. :) Well done shruti and well done Facebook :) :)
ReplyDeleteNice title to start with :)Had never given a thought about how the top news or recent news were shown.Was confused at first reading NFO diagram but the explanation that followed was very clear :)was good to know there are different kinds of edges formed.all seemed to fit in well:)Hey this algorithm is only for top news right.Was curious whether the most recent news would take only the time factor into account among the three:)Neatly explained:)good job:)
ReplyDeleteAlways use facebook,but never had thought how its features work..A very good topic chosen..thumbs up there..:):)
ReplyDeletevery interesting to know that inputs are taken as objects and when there is an interaction an edge is added...wow!!..:):)Affinity,weight ,time very well explained..:):)
Thnks to these components, they make our fb life more interesting by showing what we like..;):)and thnks to you for this blog..:):)
Happy facebooking..:):)
Facebooking kills time day in and out :P.Never knew it implements so many things just to give news feeds.Affinity, weight, time explained really well :).The Daigram itself explains the whole idea of news feed optimization. V inovative topic to choice!
ReplyDeleteexcellent
ReplyDeletethe need for real-time statistical analysis is eliminated by clever application of graph theoretic concepts.
this is very nice indeed. your condensing of the white paper to this concise article.
so..kudos,and keep up the good work
nice blog entry...simple and understandable.
ReplyDeleteby understandable, i meant not much of math(don't kill me sudarshan sir :-P)
but, to be frank, i felt that you could do away with the little mathematical eqns out there as the whole concept would still be communicated.:-)
now coming to news feed itself, it is a great piece of algorithm, a brilliant idea in fact. but do you people actually find it useful? i never glance at news feed and simply jump to the "fresh booty" tab. and in such a situation, it will be like a diamond in an ignorant child's hand.:-)
very nicely written....easy to understand......affinity,weight and time have been explained very well...
ReplyDeletein particular about weight,the priority of assigning weights is very well explained
over all enjoyable read and very informative.....
very well done!!
hadn't known so much about fb.. awesome to know.. thank you very much for the info.
ReplyDeleteVery nice post indeed!
ReplyDelete"..[It]displays stories based on their relevance rather than in chronological order.." was quite surprising to me. I actually went back and checked on my FB homepage and it is true. :)
And yes knowledge of how the 'Top News' works greatly help brands put themselves on most people's FB homepages.
thank you all :-)
ReplyDeleteFacebook makeover...
Blue Corner :-)
Top news and most recent combined together!
Can easily make out that they still use the same 'Edge ranking' algorithm.
:-)) LoL
ReplyDeleteWonder what is the weighting criteria for TN and MR