EmailSum

该数据集名为EmailSum，包含了2549个电子邮件线索的摘要，这些线索涵盖了广泛的主题，每个线索包含3至10封电子邮件。摘要分为简短（不超过30个单词）和长篇（不超过100个单词）两种形式，均由人类通过亚马逊土耳其机器人（Amazon Mechanical Turk）进行标注。为确保质量，该数据集已经过筛选，并包含了匿名的电子邮件线索。规模上，该数据集包含了2549个带有摘要的电子邮件线索，其任务是对电子邮件线索进行摘要概括。

This dataset, named EmailSum, contains 2549 email threads covering a wide array of topics. Each thread consists of 3 to 10 individual emails. Two types of summaries are available: short summaries (no more than 30 words) and long summaries (no more than 100 words), all of which were manually annotated by human workers through Amazon Mechanical Turk. To ensure data quality, the dataset has been screened and all included email threads are anonymized. In total, this dataset includes 2549 email threads paired with their respective summaries, and its core task is email thread summarization.