Discovery

该数据集包含174个话语标记，每个标记至少有1万个示例，这些示例是自动发现的。每个示例大约包含61个标记，代表了具有细粒度标签的最困难任务。该数据集的规模至少为每个话语标记10万个示例，其任务是话语标记检测。

This dataset comprises 174 distinct discourse markers, with a minimum of 10,000 automatically identified examples per marker. Each example contains roughly 61 tokens, representing the most challenging task equipped with fine-grained labels. The overall scale of the dataset amounts to at least 100,000 examples per discourse marker, and its target task is discourse marker detection.