联合国数据集(UNBench)
收藏资源简介:
该数据集包含了从1994年到2024年联合国安理会公开的草案决议、投票记录和外交演讲。它由伊利诺伊理工学院等机构创建,旨在为政治科学研究中的大型语言模型的应用提供支持。数据集覆盖了安理会决策过程中的三个阶段:起草、投票和讨论,包含了与这些阶段相关的四种任务,用于评估LLM在理解政治动态和模拟政治决策方面的能力。
This dataset contains publicly available draft resolutions, voting records, and diplomatic speeches of the United Nations Security Council from 1994 to 2024. It was developed by institutions including the Illinois Institute of Technology, and is designed to support the application of Large Language Models (LLMs) in political science research. The dataset covers three stages of the Security Council's decision-making process: drafting, voting, and deliberation, and includes four tasks associated with these stages, which serve to evaluate the capacity of LLMs to comprehend political dynamics and simulate political decision-making.




