UserGrammar

该数据集包含了153个模式和50个意图，用于模拟跨多个领域（如邮件、网页搜索、音乐会预订、旅行预订、日历管理和连接对象等）的用户请求。此外，该数据集专门用于评估AidMe框架在检测意图和模式方面的性能，重点在于适应性和从用户互动中学习。该数据集的规模包括从用户请求生成的5000对训练集和1000对测试集，其任务是对用户请求中的意图和模式进行检测和识别。

This dataset includes 153 schemas and 50 intents, which is designed to simulate user requests across multiple domains such as email, web search, concert booking, travel booking, calendar management, and connected objects, etc. Furthermore, this dataset is specifically developed to evaluate the performance of the AidMe framework in intent and schema detection, with a focus on adaptability and learning from user interactions. The dataset consists of 5000 training pairs and 1000 test pairs generated from user requests, and its core task is to detect and identify intents and schemas within user requests.