CovidQA, PolicyQA, TechQA
收藏资源简介:
本研究涉及三个数据集:CovidQA、PolicyQA和TechQA。CovidQA专注于COVID-19相关的问题回答,包含2019个问题-答案对,涉及病毒传播、公共卫生干预和社会影响等主题。PolicyQA包含12,102个关于美国移民和旅行政策的问题-答案对,需要对特定政策文件进行推理以确定答案。TechQA提供1,808个关于技术支持问题的例子,涉及计算机网络、软件和硬件。这些数据集用于评估通过GPT-4生成的合成数据对机器阅读理解任务的增强效果,特别是在资源有限的情况下。
This study involves three datasets: CovidQA, PolicyQA, and TechQA. CovidQA focuses on COVID-19-related question answering, containing 2019 question-answer pairs covering topics such as viral transmission, public health interventions, and social impacts. PolicyQA comprises 12,102 question-answer pairs regarding U.S. immigration and travel policies, which requires reasoning over specific policy documents to determine the correct answers. TechQA provides 1,808 examples of technical support questions, covering computer networks, software, and hardware. These datasets are utilized to evaluate the enhancement effect of synthetic data generated via GPT-4 on machine reading comprehension tasks, especially under resource-constrained scenarios.




