Pow-Wow

Pow-Wow数据集由丰田技术学院芝加哥分校的研究者创建，专注于研究Pommerman游戏环境中的目标导向人类协作通信。该数据集记录了人类玩家与AI代理在游戏中交互的观察、动作和通信，共包含2513条通信记录。数据收集过程中，参与者通过文本聊天窗口进行沟通，确保了通信的同步性和实时性。该数据集的应用领域主要在于开发和优化多代理系统中的通信策略，以提高团队协作效率和游戏胜率。The Pow-Wow dataset was created by researchers at the Toyota Institute of Technology, Chicago, focusing on investigating goal-directed human collaborative communication within the Pommerman game environment. This dataset captures observations, actions and conversational communications from interactions between human players and AI agents during gameplay, with a total of 2513 communication transcripts. During the data collection process, participants communicated via a text chat window, ensuring the synchronicity and real-time performance of the communications. The primary application scenarios of this dataset center on developing and optimizing communication strategies in multi-agent systems to improve team collaboration efficiency and in-game win rates.