opendebate

### **Introduction**

OpenDebate is a platform where renowned speakers from diverse backgrounds—including politicians, journalists, and spokespersons—gather to discuss pressing societal and geopolitical motions. Inspired by the debates held at Oxford Union, Cambridge, and Imperial College, OpenDebate combines elements of competitive debating and Model United Nations (MUN) formats. The platform is structured to encourage logical reasoning over aggressive tactics, with an adjudicator facilitating smooth discussions.

Unlike traditional debates where speakers may adopt more aggressive stances to defend their positions, OpenDebate prioritizes thoughtful discourse based on sound reasoning. Participants are less inclined to employ rhetorical strategies such as doublespeak, auxiliary arguments, or exaggerated examples. Instead, the focus is on identifying logical fallacies, exploring opposing viewpoints, and fostering constructive dialogue.

The OpenDebate dataset is an extension of the **LAION-Debate** dataset, which introduced the world's first competitive debate corpus. This dataset includes detailed transcripts of OpenDebate discussions, providing a valuable resource for researchers studying argumentation, dialogue systems, and human-AI interactions. While the transcripts are openly available for research purposes, the corresponding audio files are exclusively posted on this platform.

#### **Acknowledgement**

We are deeply grateful to the **speakers, adjudicators, and contributors** who participated in OpenDebate discussions and shared their insights on key societal and geopolitical issues. Their willingness to engage in meaningful conversations has made this dataset possible.

We also extend our thanks to the **academic and research communities** for recognizing the importance of open datasets in advancing research on dialogue systems, natural language processing, and human-AI interactions.

##### **Ethical Note**

The OpenDebate dataset is released under **EU research laws**, permitting the sharing of materials for **research purposes only**. We expect users to adhere to these guidelines and ensure responsible and ethical use of the dataset in research and training corpora.

If accessing this dataset from outside the EU, users should ensure compliance with their **local laws** regarding dataset usage.

We strongly encourage researchers to use this dataset **ethically and responsibly**, with a focus on promoting **positive societal impact**. Misuse of this dataset to develop harmful, unethical, or biased systems is strictly discouraged. Instead, it is our hope that OpenDebate will serve as a valuable resource for advancing **fair, transparent, and human-centered AI systems**.

### **引言**

开放辩论（OpenDebate）是一个汇聚各界知名人士的平台，参与者涵盖政客、记者与发言人，共同探讨紧迫的社会与地缘政治议题。该平台受牛津辩论社、剑桥大学及帝国理工学院辩论活动的启发，融合了竞技辩论与模拟联合国（Model United Nations，MUN）的形式特点。平台的架构设计旨在鼓励以逻辑推理而非攻击性策略展开交流，并由评审员引导讨论有序进行。

与传统辩论中发言者为捍卫立场往往采取激进姿态不同，开放辩论优先推崇基于严谨推理的深度思辨交流。参与者较少使用欺瞒性话术、辅助论证或夸张例证等修辞策略，转而将重点放在识别逻辑谬误、探讨对立观点以及构建建设性对话之上。

开放辩论数据集是**LAION-Debate**数据集的延伸——后者是全球首个竞技辩论语料库。该数据集收录了开放辩论讨论的完整逐字记录，为研究论辩学、对话系统以及人机交互的研究者提供了宝贵资源。尽管文本记录可公开用于研究，但对应的音频文件仅可在本平台获取。

#### **致谢**

我们谨向参与开放辩论讨论、并就关键社会与地缘政治议题分享见解的**发言者、评审员与贡献者**致以诚挚谢意。正是他们愿意开展富有意义的对话，才促成了本数据集的诞生。

我们同时感谢**学术界与研究界**对开放数据集在推动对话系统、自然语言处理以及人机交互研究方面的重要价值的认可。

##### **伦理说明**

开放辩论数据集依据**欧盟研究相关法规**发布，仅允许为研究目的分享相关材料。我们期望用户遵守上述准则，确保在研究与训练语料库构建中以负责任且符合伦理的方式使用本数据集。

若用户从欧盟境外获取本数据集，请确保遵守自身所在地区关于数据集使用的**当地法律法规**。

我们强烈鼓励研究人员以符合伦理且负责任的方式使用本数据集，致力于推动产生**积极的社会影响**。严禁将本数据集用于开发有害、不道德或带有偏见的系统。我们期望开放辩论能够成为推动构建**公平、透明且以人类为中心的人工智能系统**的宝贵资源。

搜集汇总

背景与挑战

背景概述

OpenDebate数据集记录了来自不同背景的演讲者关于社会和地缘政治议题的辩论，强调逻辑推理和建设性对话。该数据集扩展了LAION-Debate数据集，提供详细的辩论记录，适用于研究论证、对话系统及人类-AI互动。数据集仅限研究使用，需遵守欧盟研究法律和当地法规。