PolicyQA

PolicyQA是由加利福尼亚大学洛杉矶分校创建的一个阅读理解风格的问题回答数据集，包含25,017个从115个网站隐私政策中提取的例子。该数据集提供了714个人工标注的问题，覆盖了广泛的隐私实践。数据集的创建过程涉及从现有的OPP-115语料库中提取并人工标注问题。PolicyQA主要应用于隐私政策领域，旨在帮助用户快速找到与其相关的隐私信息，解决隐私政策阅读困难的问题。PolicyQA is a reading comprehension-style question answering dataset developed by the University of California, Los Angeles. It contains 25,017 examples extracted from the privacy policies of 115 websites, and provides 714 manually annotated questions covering a broad spectrum of privacy practices. The dataset is constructed by extracting and manually annotating questions from the existing OPP-115 corpus. Primarily applied in the privacy policy domain, PolicyQA aims to help users quickly find relevant privacy information and address the challenges of reading privacy policies.

搜集汇总

背景与挑战

背景概述

PolicyQA是一个专注于隐私政策文本的阅读理解数据集，旨在支持自然语言处理中的问答任务。该数据集包含用于微调BERT模型的源代码，并在EMNLP 2020论文中报告了相关实验结果，验证集F1分数为59.2，测试集F1分数为55.3。