FIND

该数据集包含了来自9个金融领域的实际对话记录，共有290万条真实世界的语句，这些数据源自一个金融虚拟个人助手。此外，还提供了一个包含1500条标注语句的测试集，其标注者之间的一致性大于0.75。规模上，数据集包含了290万条语句，任务是对话意图和槽位诱导。This dataset comprises real conversational records across 9 financial domains, consisting of 2.9 million real-world utterances sourced from a financial virtual personal assistant. Additionally, a test set containing 1500 annotated utterances is provided, with an inter-annotator agreement greater than 0.75. With a total scale of 2.9 million utterances, the dataset focuses on the tasks of dialogue intent and slot induction.