E-commerce Dark Pattern Dataset

E-commerce Dark Pattern Dataset是由日本早稻田大学构建的一个数据集，包含1,178个暗模式文本和1,178个非暗模式文本，总计2,356条数据。该数据集源自2019年Mathur等人的研究，通过从购物网站收集的文本数据，并添加了非暗模式文本以实现平衡。数据集的创建旨在通过自动检测技术识别和分类电子商务网站中的暗模式，以保护用户隐私和公平性。该数据集已用于评估多种机器学习模型，包括BERT、RoBERTa等，以提高自动检测的准确性。

E-commerce Dark Pattern Dataset is a dataset constructed by Waseda University, Japan. It contains 1,178 dark pattern texts and 1,178 non-dark pattern texts, totaling 2,356 entries. The dataset originates from a 2019 study by Mathur et al., where text data was collected from e-commerce websites, and non-dark pattern texts were added to balance the dataset. The dataset was developed to identify and classify dark patterns on e-commerce websites via automated detection technologies, aiming to protect user privacy and fairness. This dataset has been used to evaluate various machine learning models including BERT, RoBERTa, and others, to improve the accuracy of automated dark pattern detection.