SAID (Social media AI Detection)
收藏资源简介:
SAID是由上海财经大学开发的一个新型基准数据集,用于评估AI文本检测模型在真实社交媒体平台上的能力。该数据集包含了来自知乎和Quora等流行社交媒体平台的真实AI生成文本,旨在解决AI生成文本在社交媒体上的滥用问题。SAID通过收集真实的AI生成内容,提供了一个更真实和具有挑战性的评估环境。数据集的创建过程中,利用了Zhihu平台自动检测标签来确定AI生成文本的真实性。SAID的应用领域主要集中在提高AI生成文本的检测准确性,以及探索人类识别AI生成文本的能力。
SAID is a novel benchmark dataset developed by Shanghai University of Finance and Economics, tailored for evaluating the performance of AI text detection models on real-world social media platforms. This dataset includes authentic AI-generated texts sourced from mainstream social media platforms such as Zhihu and Quora, with the goal of addressing the abuse of AI-generated texts on social media. By collecting real-world AI-generated content, SAID provides a more realistic and challenging evaluation environment. During the dataset construction process, the automatic detection tags provided by the Zhihu platform were leveraged to confirm the authenticity of the AI-generated texts. The primary application scopes of SAID focus on enhancing the detection accuracy of AI-generated texts and investigating human ability to identify such texts.

- 1Who Said That? Benchmarking Social Media AI Detection上海财经大学 · 2023年



