GECO

该数据集是一个经过性别控制的文本数据集，其中相同语义的句子分别以男性和女性形式出现，为性别分类任务提供了真实的“世界解释”。此外，该数据集被用于评估解释性能的XAI方法的正确性，并研究模型解释中可能存在的偏见。该数据集的任务包括性别分类以及对使用XAI方法的解释性能进行评估。

This dataset is a gender-controlled text dataset where sentences with identical semantics are presented in both male and female grammatical forms, providing authentic "world explanations" for gender classification tasks. Additionally, this dataset is utilized to evaluate the correctness of XAI methods for measuring explanation performance, as well as to investigate potential biases present in model explanations. The tasks encompassed by this dataset include gender classification and the evaluation of explanation performance using XAI methods.