TikTok reveals details of how its algorithm works

Post by pb7 » Thu, Sep 10, 2020, 3:17 PM UTC

disgruntledphd2 wrote: post #45

jakear wrote: post #8

It's a (seemingly) pretty simple matter of keeping track of how long someone looks at a video, and optimizing to show them videos they'll spend a long time looking at. It's pretty powerful too, I spend much more time looking at TikTok than any other digital content aggregator (to the extent I had to delete it from my phone). They also do some things like sprinkle in random fresh videos, potentially unrelated to your…

I actually think it's got much more to do with the training data. Given that you are watching a video, you need to either swipe away from it, or finish watching it. This provides either a 1 or 0 for the video classification model. The important contrast here is with FB/IG feed where you can scroll aimlessly without engaging, leaving you with perhaps 1 engagement out of 10 (or whatever). The attached doc suggests that…

Post by jordache » Thu, Sep 10, 2020, 3:20 PM UTC

wombatmobile wrote: post #46

jordache wrote: post #36

don't see how tiktok is so game changing? How long can one look at random dance videos? What demographic would maintain sustained interest over years? flavor of the month/year?

How many minutes have you spent watching TikTok?

Post by ramraj07 » Thu, Sep 10, 2020, 3:27 PM UTC

xnx wrote: post #9

ChefboyOG wrote: post #4

I don't see any details about its algorithm, just that they use a recommendation engine? Things like this aren't proprietary info, they're just how recommendation engines work: "Once TikTok collects enough data about the user, the app is able to map a user's preferences in relation to similar users and group them into "clusters." Simultaneously, it also groups videos into "clusters" based on similar themes, like "bas…

More data beats smarter algorithms any day, and TikTok gets a lot of data because its videos are so short and interaction rate so high. There are tons of signals it can use as inputs: How much of a video did you let play before swiping next? How many times did you let the video loop? Did you like? Did you comment? Did you like a comment? Did you click through to the profile? Did you view other videos from the profile…

Post by btilly » Thu, Sep 10, 2020, 3:33 PM UTC

xnx wrote: post #18

gcmac wrote: post #15

Earlier quoted context omitted.

I’d say it’s more likely they have super advanced/clever ways of doing the latter. The algorithm could be a simple dot product and the result could be great or terrible depending on how good the feature extraction is. Pulling useful features out of videos is no small task. The fact that everyone raves about how good the recommendations are indicates to me that this is where their innovation lies.

There's so much good meta-data (likes, comments, duration, sound used, views, like/view ratio, skips, loops, subscribes, etc.) that I'd be surprised if they were digging into the contents of the video at all right now.

Post by jakear » Thu, Sep 10, 2020, 3:38 PM UTC

disgruntledphd2 wrote: post #45

jakear wrote: post #8

It's a (seemingly) pretty simple matter of keeping track of how long someone looks at a video, and optimizing to show them videos they'll spend a long time looking at. It's pretty powerful too, I spend much more time looking at TikTok than any other digital content aggregator (to the extent I had to delete it from my phone). They also do some things like sprinkle in random fresh videos, potentially unrelated to your…

I actually think it's got much more to do with the training data. Given that you are watching a video, you need to either swipe away from it, or finish watching it. This provides either a 1 or 0 for the video classification model. The important contrast here is with FB/IG feed where you can scroll aimlessly without engaging, leaving you with perhaps 1 engagement out of 10 (or whatever). The attached doc suggests that…