EDTT: Pro-Eating Disorder TikTok Dataset Analyzed
A TikTok (pro-eating disorder content) dataset of 85 videos and photo posts annotated using the EDE clinical framework, collected by Shaveet, Viranda, and Choudhury at Cornell University. Each record contains content type (video/photos), human-written content summary, EDE subscale and diagnostic item annotations, visual and audio element labels, audio tempo/valence/arousal scores, harmonic ratio, and mode. EDTT is the first clinically-grounded dataset of pro-eating-disorder TikTok content, with 47 videos and 38 photo posts annotated using the Eating Disorder Examination framework by Cornell researchers. Suitable for eating disorder content detection, harmful content moderation, multimodal annotation research, TikTok platform analysis, clinical NLP, audio-visual content analysis.