INQUISITIVE

该数据集名为Inquisitive，由人类标注者根据新闻文章生成的问题组成，并标注了问题类型，以便于进行类型控制的问答生成。该数据集涵盖了多种问题类型，如解释、详述、背景、定义、实例化和前瞻性等，并由专家标注者进行了注释。在规模上，该数据集包含了训练集15,897个实例，测试集1,885个实例，以及验证集1,984个实例。其任务旨在进行问题生成与分类。

This dataset, named Inquisitive, consists of questions generated by human annotators based on news articles, with their corresponding question types annotated to enable type-controlled question generation. It covers a variety of question types including explanation, elaboration, background, definition, instantiation, and prospectivity, and was annotated by expert annotators. In terms of scale, the dataset contains 15,897 training instances, 1,885 test instances, and 1,984 validation instances. The core task of this dataset is question generation and classification.