If you didn’t know about TikTok before the coronavirus, chances are you have it on your phone today. Get this: recent data shows this video-sharing app hit 2 billion downloads worldwide during the COVID-19 pandemic.
As TikTok’s popularity continues to soar, more people are interested in how the company goes about selecting and screening content. For instance, many users want to know how TikTok executives determine what appears on their personalized “For You” playlist.
Although TikTok has traditionally been tight-lipped about its technology, it recently published a fascinating blog post on this very topic. According to company leaders, TikTok primarily uses data related to personal preferences and video engagement to determine what content appears on your “For Me” feed.
The “For Me” algorithm takes dozens of features into account, but the most significant factors appear to be:
- How long you watch a selected video.
- Whether you share the content with friends.
- If you choose to follow the creator.
The TikTok algorithm also takes into account whatever preferences you set when you start using the app. You could also significantly impact the TikTok algorithm by liking or disliking videos.
Most of the other data points TikTok uses to create your “For Me” page don’t influence what videos appear all that much. For instance, the company does take into account where you live, what language you speak, and what phone you’re using, but they aren’t determining features.
Interestingly, the “For Me” algorithm doesn’t seem to be too concerned with whatever is currently trending. Indeed, many of the videos suggested on “For Me” filters could be a few months old. Once again, the company highlights that it’s far more interested in user preferences than promoting viral content when making suggestions.
Later in the post, TikTok said it was aware recommendation filters have a habit of creating what it termed “filter bubbles.” In other words, since algorithms only suggest what users are interested in, it could deprive them of access to alternative viewpoints and exciting new artists.
To help deal with this issue, TikTok says it sometimes suggests videos with hashtags or topics unrelated to previous search history. By including these random recommendations, users can see content they might not otherwise search out.
Another issue TikTok addressed in this post was whether the algorithm took visual info into account when recommending different videos. Apparently, the company only scans videos to ensure they meet their safety standards (e.g., no abusive content or illegal activities). You could find a full list of posts that violate TikTok’s policies on this page.
While this blog didn’t answer all the specifics of TikTok’s “For Me” algorithm, it was a significant glimpse into how the Chinese-based company operates. We can expect to learn more about the TikTok platform after the company opens its first “Transparency Center” in Los Angeles.
If you would like to read the TikTok blog post in full, please visit this official webpage.
- Twitch Reaches Out to Creators After Dozens of DMCA-Related Deletions – November 2020 - December 15, 2020
- Wikipedia Wages War On “Fake News” – New Content Restrictions Announced for 2020 Election Page - October 28, 2020
- July 2020 Pinterest Algorithm Update – Home Feed to get More Diversity & Video Content - July 12, 2020