Time-Sensitive QA (TimeQA)
收藏资源简介:
TimeQA数据集由加州大学圣巴巴拉分校计算机科学系创建,专注于时间敏感问题的回答。该数据集包含20,000个问答对,涉及5,500个时间演化事实和70种关系,旨在推动模型在时间理解和推理方面的能力。数据集通过从WikiData挖掘时间演化事实,并结合众包工作者的验证和校准,生成多样化的问答对。TimeQA数据集的应用领域包括提升自然语言处理模型对时间变化的敏感性,特别是在长文档阅读理解中。
The TimeQA dataset was created by the Department of Computer Science at the University of California, Santa Barbara, and focuses on answering time-sensitive questions. Comprising 20,000 question-answer pairs, the dataset covers 5,500 temporally evolving facts and 70 types of relations, aiming to advance models' capabilities in temporal understanding and reasoning. The dataset is constructed by extracting temporally evolving facts from Wikidata, followed by validation and calibration via crowdworkers to generate diverse question-answer pairs. Applications of the TimeQA dataset include enhancing the temporal sensitivity of natural language processing (NLP) models, particularly in long-document reading comprehension tasks.




