Chat-Doctor

该数据集由一个在线医疗论坛的聊天历史构建而成，患者可以在该论坛上提出与医疗相关的问题。它包含了患者与医生之间的互动，内容涉及各种健康问题和医疗指导。此外，该数据集被用于训练SudoLM模型，以控制对敏感医疗信息的访问，其中20%的数据被留作评估之用。该数据集规模大约包含7,320个样本，覆盖了广泛的健康话题。其任务是基于用户凭证的医疗问答访问控制。This dataset is constructed from chat histories of an online medical forum, where patients can submit medical-related questions. It includes interactions between patients and physicians, covering a wide array of health issues and medical guidance. Furthermore, this dataset is employed to train the SudoLM model for regulating access to sensitive medical information, with 20% of the data reserved for evaluation. The dataset comprises approximately 7,320 samples spanning a broad spectrum of health topics, and its targeted task is medical question-answering access control based on user credentials.