ESTER

ESTER是由南加州大学创建的一个全面的机器阅读理解（MRC）数据集，专注于事件语义关系推理。该数据集包含超过6000个问题，捕捉了10100个事件关系对，用于推理五种最常见的事件语义关系。数据集通过自然语言查询来推理事件关系，旨在测试机器对事件之间因果、子事件、共指、条件和反事实关系的理解能力。ESTER的应用领域包括提高机器对叙事文本的理解能力，特别是在理解事件之间的复杂关系方面，为MRC领域的研究提供了一个具有挑战性的基准。

ESTER is a comprehensive machine reading comprehension (MRC) dataset created by the University of Southern California, which focuses on event semantic relation reasoning. It contains over 6,000 questions and covers 10,100 event relation pairs, designed for reasoning over the five most common types of event semantic relations. The dataset uses natural language queries to infer event relations, aiming to evaluate machines' ability to understand causal, sub-event, coreferential, conditional, and counterfactual relationships between events. Application areas of ESTER include enhancing machines' understanding of narrative texts, especially in comprehending the complex inter-event relationships, and it provides a challenging benchmark for research in the MRC domain.