ICLEv3

ICLEv3数据集是由汉诺威莱布尼兹大学创建的，用于研究学习者论证中的逻辑缺陷。该数据集包含40,089个论证实例，通过自动删除高质论证文本中的中心论点单元（ADUs）来模拟学习者可能遗漏的逻辑环节。数据集的创建过程涉及精细的ADUs删除策略，确保剩余文本的自然性，同时降低论证的质量，以反映学习者论证中常见的问题。此数据集主要用于研究论证缺陷的自动检测和重建，旨在帮助学习者提高论证质量，解决论证中逻辑不完整的问题。The ICLEv3 dataset was created by Leibniz University Hannover to investigate logical flaws in learners' arguments. This dataset comprises 40,089 argumentative instances, which are generated by automatically removing Argumentative Discourse Units (ADUs) from high-quality argumentative texts to simulate the logical gaps that learners may omit in their reasoning. The dataset construction adopts meticulous ADU removal strategies, ensuring the naturalness of the remaining text while reducing the quality of the argumentation to reflect the common issues present in learners' arguments. This dataset is primarily utilized for research on automatic detection and reconstruction of argumentative flaws, aiming to help learners improve their argumentative quality and address the problem of logically incomplete arguments.