Body like an idol: K-pop fitspiration on Tumblr – an analysis of texts and images

Authors

• Linda Achilles Universität Hildesheim
• Thomas Mandl Department of Information Science and Natural Language Processing, University of Hildesheim, Hildesheim, Germany
• Christa Womser-Hacker Department of Information Science and Natural Language Processing, University of Hildesheim, Hildesheim, Germany

DOI: https://doi.org/10.23996/fjhw.126897

Keywords:

social media, data mining, feeding and eating disorders, computer-assisted image analysis, exercise, psycholinguistics

Abstract

Eating disorders are a major health issue in societies today which oftentimes remain untreated. In social media, such as Tumblr, people build communities to exchange information and connect to each other using specific hashtags. Some of these trends which emerge around these hashtags, are related to eating disorders. This study in information science addresses how inspiration for fitness (Fitspiration) inspired by music fandom (in particular K-pop) can be characterized on Tumblr by automatically analyzing text and images of posts. Images are evaluated based on their colorfulness and emotional measures, texts undergo a sentiment and readability analysis, as well as an evaluation of their psycho-linguistic features. Furthermore, a qualitative content analysis of K-pop Fitspiration posts (n=119) is performed and they are compared to the K-pop Thinspiration posts, regular Thinspiration and control group posts. Results reveal, that K-pop Fitspiration posts are oftentimes more similar to posts from the control group than to Thinspiration posts, but that they also share psycho-linguistic features with posts of eating disordered users.