In an information-seeking conversation, a user may ask questions that are under-specified or unanswerable. An ideal agent would interact by initiating different response types according to the available knowledge sources. However, most current studies either fail to or artificially incorporate such agent-side initiative. This work presents InSCIt, a dataset for Information-Seeking Conversations with mixed-initiative Interactions. It contains 4.7K user-agent turns from 805 human-human conversations where the agent searches over Wikipedia and either directly answers, asks for clarification, or provides relevant information to address user queries. The data supports two subtasks, evidence passage identification and response generation, as well as a human evaluation protocol to assess model performance. We report results of two systems based on state-of-the-art models of conversational knowledge identification and open-domain question answering. Both systems significantly underperform humans, suggesting ample room for improvement in future studies.

翻译：在信息寻求对话中，用户可能提出定义不明确或无法回答的问题。理想化的智能体会根据可用的知识源，通过发起不同类型的回应进行交互。然而，当前多数研究要么未能实现此类智能体方的主动性，要么是人为地引入这种主动性。本文提出了InSCIt，一个面向混合主动交互的信息寻求对话数据集。该数据集包含来自805段人-人对话的4700个用户-智能体轮次，其中智能体在维基百科上搜索信息，并采用直接回答、请求澄清或提供相关信息中的一种方式来响应用户查询。该数据集支持两个子任务：证据段落识别和回应生成，以及一个用于评估模型性能的人工评估协议。我们报告了基于当前最先进的对话知识识别和开放域问答模型的两个系统测试结果。两个系统的表现均显著低于人类，表明未来研究仍有很大改进空间。