SexTok: Multi-Modal Dataset for Sex Education vs. Suggestive Content Classification

Paper @ ACL Findings 2023 | 🏆 Leaderboard | Data | Workshop on Online Harms and Abuse | Video

---

Overview

It’s not Sexually Suggestive; It’s Educative | Separating Sex Education from Suggestive Content on TikTok videos (George & Surdeanu, Findings 2023)

SexTok is a multi-modal dataset of 1000 TikTok videos addressing the challenge of distinguishing between sexually suggestive content and sex education videos. The dataset includes: three class labels: Sexually Suggestive (20%), Sex Education (20%), Others (60%), audio transcriptions using OpenAI Whisper and gender expression annotations for bias evaluation

Example

Leaderboard

Performance on SexTok test set, sorted by Macro F1:

Model	Accuracy	Macro F1	Source
🏆 Consensus-Aware Balance Learning (Zhou et al.)	86%	84%	Zhou et al.
SlowFast	80%	76%	Zhou et al.
ResNet	77%	67%	Zhou et al.
TimeSformer	75%	68%	Zhou et al.
Uniformer	74%	68%	Zhou et al.
VideoMAE	70%	61%	George et al.
BERT (Transcription)	68%	64%	George et al.

📧If you’ve used SexTok in your work, and would like to be added to the list above, please email us! Contact Info below

Citation

@inproceedings{george-surdeanu-2023-sexually,
   title = "It{'}s not Sexually Suggestive; It{'}s Educative | Separating Sex Education from Suggestive Content on {T}ik{T}ok videos",
   author = "George, Enfa  and
     Surdeanu, Mihai",
   editor = "Rogers, Anna  and
     Boyd-Graber, Jordan  and
     Okazaki, Naoaki",
   booktitle = "Findings of the Association for Computational Linguistics: ACL 2023",
   month = jul,
   year = "2023",
   address = "Toronto, Canada",
   publisher = "Association for Computational Linguistics",
   url = "https://aclanthology.org/2023.findings-acl.365/",
   doi = "10.18653/v1/2023.findings-acl.365",
   pages = "5904--5915",
}

Contact

enfafane <\a> gmail.com