Clustering Journalistic TikTok Videos
Lion Wedel, Anna-Theresa Mayer · 2025
Given the growing importance of visual social media platforms, such as Instagram, Snapchat, YouTube, and TikTok, for news consumption and the distribution of journalistic content, it has become crucial for journalism scholars to conduct the (quantitative) analysis of visual data in large quantities by employing automation. This chapter guides readers through the analysis of video data using unsupervised machine learning methods, specifically the clustering of videos, and elaborates on the methodological decisions made in this process. Using a dataset of journalistic TikTok videos from 2023, the chapter covers (1) the feature extraction from videos via transfer learning, including the required frame sampling and pre-processing steps, and (2) the application of unsupervised machine learning (e.g., clustering algorithms) to the video representations, including the quantitative and qualitative interpretation and validation of the clusters.