This project aims to integrate the monitoring infrastructure developed by Tracking Exposed and Junkipedia to monitor the Tiktok recommendation algorithm. This combined experimental pipeline will be used to establish if the recommender system differ.
NGI-ATLANTIC:OC05-387
When analysing social media platforms, their proprietary recommendation solutions are considered to be “black boxes”. The complexity and opacity of those algorithms requires to make an empirical comparison between the suggested contents on the Tiktok Web interface and its Mobile application, in order to answer our main research question.
To do this we will deploy the Junkipedia Mobile analysis solution combined with TkTrex, a software developed by Tracking Exposed. As a final deliverable of this project a technical report will be produced. It will consider which interface (web or mobile) is more prone to give biased content recommendations to users and, if this difference exists, we will consider which platform should be more thoroughly analysed.
Our overall goal is to create an open-source and accessible infrastructure to monitor the ongoing censorship trends on TikTok. The end goal is to create an application where users, journalists and NGOs can monitor a specific topic on TikTok.
Marc Faddoul, (AI Forensics)
Not available yet
This experiment is currently underway.
Country: France United States
NGI Project: NGI atlantic
Status: Early research demo
Category: Measurement, monitoring, analysis and abuse handlingVertical use cases, improving search and discovery and community building