Google's Taskmaster-1 Dataset
收藏arXiv2025-09-30 收录
下载链接:
https://github.com/google-research-datasets/taskmaster
下载链接
链接失效反馈官方服务:
资源简介:
该数据集通过“巫师奥兹”方法收集了用户与自主对话系统的互动数据,其中包含了标记好的意图和对话中的槽位信息。此外,利用谷歌文本转语音API和CLUSTERGEN,将翻译后的对话文本合成为音频。该数据集涵盖了3243条话语,分布在6个意图类别中,其任务是针对印度语系的语言进行意图识别。
This dataset collects interaction data between users and autonomous dialogue systems via the Wizard of Oz method, which includes labeled intent and slot information within the dialogues. Additionally, the translated dialogue texts are synthesized into audio using the Google Text-to-Speech API and CLUSTERGEN. Comprising 3243 utterances distributed across 6 intent categories, this dataset is developed for intent recognition tasks targeting Indo-Aryan languages.
提供机构:
Google



