PRESTO

PRESTO数据集是由谷歌公司创建的，包含超过55万条多语言真实任务导向对话，旨在解决现有数据集在捕捉用户痛点方面的不足。该数据集包含多种挑战，如不流畅性、代码切换和用户修正，这些都是在自然语言理解任务中常见的现象。PRESTO数据集不仅包含真实的多语言对话，还提供了结构化上下文，如用户的联系人和列表，以增强对话的真实性。数据集的应用领域广泛，主要用于研究和改进虚拟助手的对话解析能力。

The PRESTO dataset was created by Google, containing over 550,000 multilingual real task-oriented dialogues. It aims to address the limitations of existing datasets in capturing user pain points. This dataset encompasses multiple challenges prevalent in natural language understanding (NLU) tasks, including disfluencies, code-switching, and user self-corrections. In addition to authentic multilingual dialogues, the PRESTO dataset also provides structured contextual information such as users’ contacts and lists to enhance the realism of the dialogues. The dataset has a wide range of application scenarios, and is primarily used for researching and improving the dialogue parsing capabilities of virtual assistants.