TinySocial

TinySocial数据集由范德堡大学开发，旨在为中学生提供社交和心智理论（ToM）推理的学习材料。该数据集包含约100个视频片段，每个片段附带6-12个多选题，内容涉及情感识别、语言理解等。数据集的创建过程包括手动搜索Youtube视频、内部问题编写和通过Amazon Mechanical Turk进行众包问题编写。TinySocial数据集主要用于教育游戏，帮助自闭症谱系的中学生提高社交和ToM推理能力，同时也可能用于测试人工智能的社交推理能力。

The TinySocial dataset was developed by Vanderbilt University, aiming to provide learning materials for middle school students focused on social and Theory of Mind (ToM) reasoning. This dataset contains approximately 100 video clips, each accompanied by 6 to 12 multiple-choice questions covering topics such as emotion recognition and language comprehension. The development process of the TinySocial dataset includes manually searching YouTube videos, in-house question writing, and crowdsourced question creation via Amazon Mechanical Turk. The TinySocial dataset is mainly used for educational games to help middle school students with autism spectrum disorder improve their social and ToM reasoning abilities, and it can also be employed to test the social reasoning capabilities of artificial intelligence.