A recent study in the U.S. and Europe presents an in-depth analysis of TikTok’s recommendation system for videos based on user preferences and likes.

Research findings

Because of the European Union’s data protection laws, researchers from Boston University and the University of Washington were able to examine user data from TikTok. According to their analysis, between 30 and 50 percent of the first 1,000 videos that are suggested to newly created TikTok accounts correspond with the known preferences of the users.

How the TikTok algorithm works

TikTok’s algorithm system takes into account multiple variables to customize videos to users’ preferences. Similarities between liked videos, creators you follow and posted content themes are explored. Surprisingly, users’ viewing times don’t seem to have a significant impact on the recommendation system, raising concerns about the platform‘s definition of a “view.”

The researchers claim that TikTok’s algorithm works in two ways: “explorative,” which introduces a variety of content to see how users respond, and “exploitative,” which shows content based on user behavior that it thinks users would find entertaining.

Diverse user experiences

Additionally, the study found that users receive recommendations that are both exploitative and exploratory to differing degrees. According to Franziska Roesner from the University of Washington, users of TikTok have varying experiences because the algorithm handles them differently based on unidentified factors.

Insights and future research

Boston University’s Karan Vombatkere noted this study as a crucial first step in comprehending TikTok’s customization strategies. Subsequent investigations will focus on examining particular elements that affect personalization and how they affect users’ experiences consuming information.

Featured image asset courtesy: TikTok