Current location - Loan Platform Complete Network - Big data management - What is the Shake Shack algorithm like?
What is the Shake Shack algorithm like?

1, machine review + artificial double review

When a video is initially uploaded, the platform will give you an initial flow, if after the initial flow, according to the rate of likes, comments, forwarding rate, to judge: the video is popular or unpopular, if the first round of judging as popular, then he will make a second spread.

When the second time to get the optimal feedback, then it will be given to recommend you greater traffic.

On the contrary, in the first wave or the nth wave, the response is not good, it is no longer recommended, without the recommendation of the platform, your video want to fire the probability is minimal, because there is no more traffic can see you. The first step of the video fire is to be seen by others, the first step of the road to death, the follow-up can only rely on friends star points of praise.

The logic of thinking behind this algorithm:? Intelligent distribution, overlay recommendation, and heat weighting.

2, superimposed recommendation

The so-called superimposed recommendation, refers to the new video will be intelligently distributed about 100vv playback, such as the amount of retweets up to 10 (for example), the algorithm will be judged to be popular content, automatically weighted for the contents of the superimposed recommendation to you 1000vv; retweets up to 100 (for example), the algorithm continues to superimpose the recommended to 10,000vv; retweets up to 1000 (for example), and then superimposed recommended to 10wvv, and so on cumulative push? So those overnight millions of playback volume of the jitterbug owner also confused than, do not know what happened to God, in fact, is the big data algorithm weighting.

Superimposed on the recommendation is of course the comprehensive weight of the content as an assessment standard, the key indicators of the comprehensive weight: the completion of the broadcast rate, the number of likes, the number of comments, the number of retweets, and the weight of each echelon varies, when a certain amount of grades, it is a combination of big data algorithms and manual operation of the mechanism.

3, heat weighting

Real brush nearly a hundred fire jitterbugs, found that all the overnight fire video, and jitterbugs recommended board video, play more than a million level, comprehensive data (finish broadcast rate, likes, comments, retweets) without exception are very good.

Extended information:

1: Improve your own information, the more complete the better. Including avatar, nickname, cell phone, microblogging, microblogging, headlines, etc., the more detailed the better. Because it is a double review by machine and man, once the machine carries out the review, it will carry out a lot of poor quality elimination.

2, the video needs to have highlights. The video is only 15 seconds, in this short 15 seconds, there is no highlights, no twist, people are not going to have any interaction with you, and there is a shielding function, once the user has shielded you, this is a very serious matter, because the later will not give the user to carry out the recommendation of your short video;