Stanford Multi-Turn Dialogue
收藏arXiv2025-09-30 收录
下载链接：
https://nlp.stanford.edu/blog/a-new-multi-turn-multi-domain-task-oriented-dialogue-dataset/
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含了3,031个关于车载助手领域的对话，这些对话能够自动回应并基于一个只有车载助手知晓的个性化知识库（KB）。该知识库的条目编码了满足驾驶员提出查询的信息，包括三种类型的知识库：活动日程、每周天气预报以及兴趣点导航信息。该数据集的词汇量达到了14,000个。规模上，该数据集包含了3,031个对话，其任务是对话回应生成。
This dataset includes 3,031 dialogues in the in-vehicle assistant domain, which can automatically generate responses based on a personalized knowledge base (KB) exclusively known to the in-vehicle assistant. Entries within this KB encode information that addresses queries from drivers, covering three types of knowledge bases: activity schedules, weekly weather forecasts, and point-of-interest (POI) navigation information. The dataset has a vocabulary size of 14,000. In terms of scale, this dataset contains 3,031 dialogues, and its task is dialogue response generation.
提供机构：
Stanford University
