Press release
What makes TikTok tick? BMBF Joint Project DataSkop analyses the Platform's For You Feed
As part of the Human-Technology Interaction for Digital Sovereignty funding priority, a cooperative data donation project is being launched today to investigate the personalisation of the For You feed on the TikTok platform and enable users to experience their usage behaviour with the help of visualisations even before they donate data.
What do TikTok users see in their For You feed? Which niches are they directed to? This is what DataSkop's second data donation project aims to find out in the coming months. Users can voluntarily provide their data to the project until the end of March 2023. These data sets will be used to create a concrete picture of how the platform's recommendation algorithm works.
How does the data donation process work?
Users download the open source software at http://dataskop.net which requests their GDPR data for them. This is private usage data that platforms are required to disclose under the General Data Protection Regulation (GDPR). As soon as this data is available, it is retrieved automatically. Users then see three visualisations of their usage behaviour, which the team from the University of Applied Sciences Potsdam, led by Prof. Constanze Langer, Professor of Visual Interface Design at the Department of Design at the University of Applied Sciences Potsdam, designed together with AlgorithmWatch. Users will then have the opportunity to donate their data to the research project.
Background to the research
Content such as disinformation and hate speech is polarising on social media platforms. This also applies to TikTok. The company has recently been the subject of heated debate, for example due to the spread of Russian propaganda, inadequate protection of minors and discriminatory content moderation. With over one billion users, TikTok is one of the most influential platforms in the world.
The DataSkop project team therefore wants to analyse the TikTok recommendation algorithm with real usage data: How and where do trends and niches emerge? Are there indications that the platform places certain content prominently in the for-you feeds? The project team is focusing its investigations on the topics of the past summer, which was characterised by war, inflation, heat disasters and coronavirus. How prevalent were these topics on TikTok? How clearly do the for-you feeds of individual users differ from one another? After the donated data has been analysed by DataSkop partners, the results will be published.
Social media companies such as Facebook, Instagram, Google and others are often aware of the harmful effects of their algorithmic systems and endeavour to prevent independent research. Data donations are one of the few ways to investigate opaque algorithms. The public has a legitimate interest in gaining access to platform data: This will help to identify potential risks to society and democracy, hold companies to account and develop effective regulatory mechanisms for platforms.
Project partners
AlgorithmWatch (project coordinator), ENS/Europa Universit?t Viadrina, Paderborn University – Institute of Educational Sciences, medialepfade.org – Association for Media Education and the University of Applied Sciences Potsdam – Interaction Design Lab at the Department of Design.
More information on the data donation software and the research project can be found at http://dataskop.net