LIQUID

LIQUID是一个自动生成列表问题回答数据集的框架，由高丽大学开发。该数据集通过从无标签的语料库中提取摘要和命名实体，生成包含多个非连续跨度的列表问题。数据集主要用于训练和评估QA模型，特别是在多跨度QA任务中。LIQUID数据集的应用领域包括一般领域和生物医学领域，旨在解决现有QA数据集中列表类型问题不足的问题。LIQUID is a framework for automatically generating list question answering datasets, developed by Korea University. It generates list questions containing multiple non-continuous spans by extracting summaries and named entities from unlabeled corpora. This dataset is primarily utilized for training and evaluating QA models, particularly in multi-span QA tasks. The application domains of the LIQUID dataset cover general and biomedical fields, aiming to address the scarcity of list-type questions in existing QA datasets.