How do YouTube recommendations work
Are there any readers who do not visit YouTube at least once a day? Service has already become an integral part of our life. It is hard to imagine that we will use something else to watch videos. YouTube offers a huge content base. In total, the service has 1.9 billion monthly active users. According to statistics, 79% of Internet users have a YouTube account. So how does Google manage to keep such a huge product running? In this article, we will look at how the YouTube suggestion algorithm works, and it's interesting, believe me.
This material will be based on the official publication of Google, which explains how YouTube recommendations algorithms based on neural networks work. Why did I decide to study this issue? The fact is that not so long ago, before going to bed, I decided to turn on the sounds of the waterfall (white noise) in order to fall asleep faster. The next evening at the same time, I noticed that the very first place in the recommendations was the same video. I turned it on again. On the third day at the same time, this video was again in the same first position. And this despite the fact that at any other time, YouTube recommends me completely different videos.
And then I finally realized that YouTube's algorithms work much more complicated than we think. At a minimum, they are able to adapt to your preferences at different times of the day. Then I decided to study how YouTube algorithms work and came across some interesting information that I am ready to share with readers.
YouTube developers faced several challenges when developing the algorithm:
A huge number of videos in various topics, which complicates the optimal selection in recommendations
High service dynamics. Hundreds to thousands of hours of videos are uploaded to YouTube every hour. The recommendation system needs to be flexible and dynamic
Variability of viewers' interests
Optimization of resources for the selection of recommendations, since the work of selection algorithms is a complex process that requires a lot of power
YouTube recommender architecture
Millions of videos are submitted to the system, and at the output it offers the same dozens of videos that get to the user on the screen in the "Recommendations" tab.
The system consists of two convolutional neural networks: "candidate generation" and "ranking" (ranking). The first network of millions of videos selects hundreds of the most suitable ones, the second neural network ranks the resulting selection from the most to the least interesting to the user. When making a sample, the system takes into account the entire user history and context. Context refers to, for example, time of day, age, gender, geographic location. Also, at the moment of creating a sample, A / B testing takes place, when, for the sake of experiment, the user is shown various samples, if any of the samples turns out to be more visible, the system self-learns and adapts to this sample. Best seo company in Lagos Nigeia s When evaluating the sample, not only the viewing time is taken into account, but also the CTR (click through rate) - the number of users who started watching the video in relation to the number of users who saw the video in the recommendations.Travel agency in Lagos
At the ranking stage, the sample is based on expected watch time, so the longer users watch a video, the higher the chance that it will be in the top recommendations. YouTube isn't just about click-through rate, as videos can be just clickbaits. The goal of the ranking neural network training is to predict the video viewing time. You can also check out this How to activate YouTube Using Youtube com. best skincare routine for dry skin
YouTube recommendations are formed from two neural networks. The first neural network is responsible for the selection of videos on the topic, the second-level neural network among the selected cuts out clickbait and uninteresting videos with low user engagement. That is why videos that watch longer, like and comment more often fall into the very first places in the recommendations, if they correspond to topics that are interesting to the user. Interesting, isn't it? Let's discuss this topic in Telegram. Buy sanitaries online, Lagos sanitary ware dealers, kitchen toilet bathroom jacuzi accessories
The system is really complex and I will not try to explain complex terms and the complete architecture of the stages of forming a selection, simply because I myself do not fully understand how it works, but it is obvious that the selection of recommendations along with the Google search is the most complex algorithm on which the best work the minds of the world.