Designing a Prototype Visualization Based on White-Box Indicators to Assist Users Identifying Misinformation on TikTok

Masterarbeit im Studiengang IT Sicherheit von Robert Cieslinski:

Designing a Prototype Visualization Based on White-Box Indicators to Assist Users Identifying Misinformation on TikTok
Abstract: Digital Fake News or misinformation in general have become a central part of online discourse and research. Detection approaches often try to maintain a certain level of transparency/interpretability to counteract unintended effects (e.g., Confirmation Bias, Backfire Effect). One way for this is applying human understandable white-box indicators to the content in question. Several indicators are already described in the literature, with some of them tailored to specific platforms such as Twitter or YouTube. For TikTok, which popularized short videos with easy sound-reuse, they may have to be reevaluated or new ones have to be developed. In this work, an initial user study with 18 participants was conducted. The resulting list of helpful and potential applicable indicators showed two prominent groups of indicators: high emotionalization in some parts of the post and missing context/sources. In the next step, a visualization prototype is developed. After prefiltering by a (simulated) classification system or a factchecker’s assessment, fitting indicators are highlighted in the post. The design is evaluated with mock data.Betreuerin: Katrin Hartwig, M.Sc.

Prüfer: Prof. Dr. Dr. Christian Reuter