coding-kelps/liaisons-claim-stance-sample

本数据集是通过处理IBM Debater Claim Stance Dataset创建的样本，旨在为资源有限的环境下进行基于关系的论点挖掘任务提供基准测试。数据集的大小已减少到大约100个条目，并对原始数据集进行了修改，以实现更平衡的立场和主题分布，并创建了一个新的“不相关”类别的论点关系。数据集包括两个分割：binary和ternary，分别包含支持和攻击关系以及支持、攻击和不相关关系。数据集的结构包括主题、父论点、子论点和关系四个特征。

The present dataset is a result of processing the IBM Debater Claim Stance Dataset to create representative samples, suitable for relation-based argument mining tasks with limited resources. The dataset includes two splits: binary and ternary, with 110 and 72 samples respectively. The features include topic, parent_argument, child_argument, and relation. The dataset has been preprocessed to add an unrelated class to balance the plurality of stances and topics. The dataset is licensed under CC BY-SA 3.0.