SexTok

该数据集是一个多模态的数据集，由抖音视频组成，这些视频被标记为性暗示、性教育内容或非以上两者，旨在解决区分这类内容所面临的挑战。数据集不仅包含了类别标签和性别表达的手动注释，而且这些注释之间有高度的共识，这一共识通过Cohen's Kappa评分得到了验证。此外，数据集还包含了转录音频，其单词错误率较低。该数据集规模包含1000个抖音视频链接，其任务是实现对抖音视频的分类，具体分为性暗示、性教育和其他类别。

This is a multimodal dataset consisting of Douyin videos, which are annotated into three categories: sexually suggestive content, sex education content, and neither of the above. This dataset is developed to address the challenges associated with distinguishing such content. It not only includes manual annotations for category labels and gender expression, but also features high inter-annotator agreement, which is validated using the Cohen's Kappa score. Additionally, the dataset contains transcribed audio with a low Word Error Rate (WER). Comprising 1000 Douyin video links, this dataset supports the task of classifying Douyin videos into the aforementioned three categories.