What does a Text Classifier Learn about Morality? An Explainable Method for Cross-Domain Comparison of Moral Rhetoric - supplemental material

Raw results and crowd task material for the paper "What does a Text Classifier Learn about Morality? An Explainable Method for Cross-Domain Comparison of Moral Rhetoric", published at ACL '23. This repository contains two main folders. First, the instructions and informed consent used to perform the crowd study described in the paper, where crowd workers are asked to compare word bubbles describing moral values in a domain. Second, the results of the pairwise comparisons performed with the seven models trained on the MFTC datasets and the comparison of the results with the crowd annotations.The code to generate the results is available at this DOI: 10.4121/1e71138c-be26-4652-971a-48a84837df8eThe seven models are available at this DOI: 10.4121/646b20e3-e24f-452d-938a-bcb6ce30913c

发表于ACL 2023的论文《文本分类器如何学习道德知识？面向道德修辞跨域比较的可解释方法》的原始实验结果与众包任务材料。本仓库包含两个核心文件夹：其一为开展论文所述众包研究时所使用的任务说明与知情同意书，该研究要求众包工作者对特定领域内描述道德价值观的词汇气泡进行比较；其二为基于MFTC数据集训练得到的7个模型所完成的成对比较结果，以及将该模型结果与众包标注结果进行比对的分析内容。生成上述实验结果的代码可通过以下DOI获取：10.4121/1e71138c-be26-4652-971a-48a84837df8e。上述7个模型可通过以下DOI获取：10.4121/646b20e3-e24f-452d-938a-bcb6ce30913c