RENOVI

RENOVI数据集由莫纳什大学创建，包含9258个多轮对话，旨在探索和修复社会规范违反。数据集分为两部分：512个人类编写的对话和8746个由ChatGPT生成的合成对话。通过质量控制协议确保数据质量，数据集用于评估大型语言模型在理解和修复社会规范违反方面的能力。RENOVI特别关注中文文化背景，提供了一个独特的视角来研究社会规范在对话中的应用和影响。The RENOVI dataset, developed by Monash University, comprises 9,258 multi-turn dialogues focused on exploring and repairing social norm violations. The dataset is split into two subsets: 512 human-authored dialogues and 8,746 synthetic dialogues generated by ChatGPT. Standardized quality control protocols are implemented to guarantee data quality. This resource is intended to assess the capabilities of Large Language Models (LLMs) in understanding and repairing social norm violations. Notably, RENOVI specifically focuses on the Chinese cultural context, providing a unique perspective for researching the application and impact of social norms in conversational interactions.

搜集汇总

背景与挑战

背景概述

RENOVI是一个包含9258个多轮对话的数据集，由莫纳什大学创建，旨在探索和修复社会规范违反，特别关注中文文化背景。数据集分为512个人类编写对话和8746个ChatGPT生成合成对话，通过质量控制确保数据质量，用于评估大型语言模型在理解和修复社会规范违反方面的能力。