MindScope

该数据集名为MindScope，旨在探索决策过程中的认知偏差。其中静态部分包含了5,170个开放性问题，覆盖了72种不同的认知偏差；而动态部分则包含了超过100个场景下的多轮对话脚本。该数据集融合了静态与动态元素，以提高对认知偏差的检测能力。用户还可以修改这些对话脚本来生成个性化的数据。总的来说，该数据集包含了5,170个问题，涉及72种认知偏差，以及超过100个对话场景，其任务旨在大型语言模型（LLMs）中检测认知偏差。

This dataset, named MindScope, is designed to explore cognitive biases in decision-making processes. Its static component contains 5,170 open-ended questions covering 72 distinct cognitive biases, while its dynamic component includes multi-turn dialogue scripts across more than 100 scenarios. This dataset integrates both static and dynamic elements to enhance the detection of cognitive biases. Users can also modify these dialogue scripts to generate personalized datasets. Overall, this dataset comprises 5,170 questions related to 72 cognitive biases and over 100 dialogue scenarios, with its core task aiming to detect cognitive biases in Large Language Models (LLMs).