TikTok publishes how its content recommendation system works

FOLLOW

• Follow

When you open the TikTok short video application, a sequence of content recorded by other users appears that the platform selects and displays in a section called For You. This main video stream is different for each user, it is personalized. From the beginning, content creators have struggled to discover what characteristics their publications must have to appear in this section, since it is the one that has the most visibility of the app . Now, the company has detailed what are the factors that contribute to a video appearing in it and how the content is weighted for each user.

Like many other platforms to recommend their content, the For You section of TikTok is organized taking into account user activity, according to the company. The app takes into account the videos you like or share, the accounts you follow, the comments you post, and the content you create to help determine your interests. Creators who want their videos to appear in this section don't have much to do here. But they can adapt the information in their posts, such as subtitles, sounds, and hashtags associated with the content.

TikTok also values ​​other less obvious data, such as language preference, country and device type, although it ensures that these factors have less weight in the recommendation system compared to others. For example, if a user watches a long video from start to finish, it is considered a strong indicator of interest. This would have more weight than the fact that the viewer and the owner of the video were from the same country.

The videos with the most views are those created by accounts with many followers. In this case, the company confirms that it is simply because that account has a larger viewer base. But he adds that neither the number of followers nor the visits of the videos are considered decisive factors in recommending them to other users. This indicates that the main influencers of TikTok are not guaranteed that their content reaches For you.

As it begins

Like any application that relies on the tastes of its users to make recommendations, TikTok has to find a formula to suggest content of interest to newcomers to the social network, for which it has no references yet. In these cases, the platform does not know what type of content an individual likes. To manage this situation, ask new users to select categories that interest them, to help adapt the initial recommendations. If users choose not to select categories, TikTok displays a general set of popular videos until it can gather more information.

TikTok also explains what it is doing to ensure that the system is not creating bubbles, that is, to prevent the videos shown to users in For You from being too homogeneous: over-serving someone's personal taste can lead to development. from limited experience. To avoid this, delete repetitive, duplicate or user-viewed content before. It also prevents two videos in a row from the same creator or with the same background sound.

Also, TikTok will add videos to your For You feed at times that don't seem relevant to your expressed interests or have accumulated a lot of likes. This is part of their attempts to add diversity, by giving users the opportunity to stumble across new content categories and new creators, and allow them to "experience new perspectives and ideas," the company says.