SLIGHT

该数据集名为SLIGHT，旨在研究对话中隐含攻击性文本的检测问题。它包含了由人类标注的推理链条，这些链条解释了如何从含糊其辞的陈述中得出攻击性的解释。数据集中的标注包括隐含攻击性陈述、明确攻击性陈述、非攻击性陈述以及推理链条。该数据集强调了语境的重要性，并包含了多样化的个人属性信息。规模上，该数据集包含了2657个示例，其任务是进行隐含攻击性文本的检测。

This dataset, named SLIGHT, is designed to study the detection of implicitly aggressive text in dialogues. It contains human-annotated reasoning chains that explain how aggressive interpretations can be derived from ambiguous statements. The annotations in the dataset cover implicitly aggressive statements, explicitly aggressive statements, non-aggressive statements, and reasoning chains. This dataset emphasizes the importance of context and includes diverse personal attribute information. In terms of scale, it contains 2657 instances, and its targeted task is the detection of implicitly aggressive text.