Twelve Angry Men dataset

该数据集记录了一起谋杀案审判中12名陪审员之间的辩论言论（发言），共将80对辩论言论进行了配对，并标注了它们之间的关系：支持或攻击。此外，该数据集还用于提取事实和推理规则，以便利用强化学习训练对话策略。该数据集的任务是基于论证的对话系统训练和评估。

This dataset records the debating utterances (speeches) among 12 jurors during a murder trial. A total of 80 pairs of such debating utterances are paired, with their relational labels (support or attack) annotated. Additionally, this dataset is utilized to extract factual information and inference rules for training dialogue policies through reinforcement learning. The task of this dataset centers on the training and evaluation of argument-based dialogue systems.