alinia/EADOP-RAG-out-of-domain

EADOP RAG域外数据集包含超过2000条人类注释的域内和域外用户消息和聊天机器人助手回应。这些消息和回应涉及当前加泰罗尼亚立法的有用信息，并包含加泰罗尼亚语、西班牙语和英语三种语言。数据集提供了用户消息、消息类型、适当性评估、适当性解释、消息语言检测、上下文、上下文相关性评估和解释等特征。

The EADOP RAG out-of-domain dataset consists of more than 2,000 human-annotated in-domain and out-of-domain user messages and assistant responses in the context of a chatbot providing helpful information about the current Catalan legislation. The dataset includes messages and responses in Catalan, Spanish, and a small fraction in English, offering features such as user messages, message types, appropriateness assessment, explanations of appropriateness, detected message language, context, relevance assessment of the context, and explanations of context relevance.