MuDoCo

该数据集名为MuDoCo，包含了6个领域的7,500个以任务为导向的多轮对话，这些对话被标注了用于核心ference解析和查询重写的信息。此外，数据集还包括了核心ference链接和重写后的查询。在数据集的构建过程中，涉及了30名标注员进行的详细标注以及质量保证步骤。规模上，该数据集包含了7,500个对话，其任务重点在于对话中的核心ference解析和查询重写。

The dataset named MuDoCo contains 7,500 task-oriented multi-turn conversations across six domains, all annotated with information for coreference resolution and query rewriting. Additionally, the dataset also provides coreference links and rewritten queries. During the construction of this dataset, detailed annotation work conducted by 30 annotators and quality assurance procedures were involved. With a total of 7,500 conversations, the core task focus of this dataset lies on coreference resolution and query rewriting within dialogues.