What are TikTok users shown when they launch the app? Which niches are they sent to? How and where do trends emerge? The -DataSkop- research collaboration, headed by the charitable organisation -AlgorithmWatch-, is seeking to find answers to this over the coming months. Users will be able to provide their data to the project from 26 January to 31 March. These datasets will then be used to gain insights into how TikTok’s recommendation algorithms work. This data donation project, funded by the Federal Ministry of Education and Research (BMBF), involves researchers from Paderborn University, the European New School of Digital Studies, the European University Viadrina, the University of Applied Sciences Potsdam and the Mediale Pfade association.
Gaining a better understanding of recommendation algorithmsSocial media platforms often contain polarising content based on false or contemptuous statements - with real consequences for users. The same applies to the video portal TikTok, which has prompted major discussion in recent times due to the dissemination of Russian propaganda, insufficient protection for young people, and discriminatory restrictions on content. With more than a billion users, TikTok is one of the most influential platforms in the world. The -DataSkop- project team is therefore seeking to examine the TikTok recommendation algorithm with the help of real usage data. -For example, we are interested in whether there are indications that the platform places particular content in prominent positions in -For You- feeds, or how strongly topics such as the war in Ukraine, inflation, heat catastrophes, and coronavirus were featured on TikTok-, explains Paderborn sub-project manager Professor Bardo Herzig. Social media companies such as Meta, Google and co. are often aware of the damaging impact of their algorithms, but prevent independent research from being conducted into them. Data donation enables researchers to examine the opaque functionalities and recommendation principles of algorithmic systems and subsequently gain insights.
Paderborn’s researchers will use the tools developed and project findings obtained for media education at the university. Analysing how usage data is collected and processed in recommendation systems forms a key cornerstone of media and data competence. -Visual approaches are hugely important for students when it comes to obscure usage data that is difficult to analyse-, explains Paderborn media educator and project member Emanuel Sarjevski.