SexTok — Separating Sex Education from Suggestive Content on TikTok
Dataset annotation for multimodal social media content moderation research.
Content moderation systems can struggle to distinguish between sexually suggestive content and legitimate educational discussions about sexual health, particularly on social media platforms.
This research project investigates this distinction using TikTok videos and introduces SexTok, a multimodal dataset for studying sexually suggestive and sex-educational content.
The dataset contains 1,000 TikTok videos annotated across categories including sexually suggestive content, sex education, and other content. The work also considers information from multiple modalities, including video, text, and audio.
My Contribution
I contributed to the annotation of the dataset, helping label and prepare examples used in the study.
The resulting dataset was used to investigate how computational models can distinguish educational sexual-health content from sexually suggestive material, highlighting challenges in automated social media content moderation.
Publication
George, Enfa and Mihai Surdeanu. “It’s not Sexually Suggestive; It’s Educative”: Separating Sex Education from Suggestive Content on TikTok videos. Findings of ACL 2023.