MineDojo Internet Knowledge Base (Wiki)

Project website: minedojo.org Paper: arxiv.org/abs/2206.08853 GitHub: github.com/MineDojo/MineDojo The Minecraft Wiki pages cover almost every aspect of the game mechanics, and supply a rich source of unstructured knowledge in multimodal tables, recipes, illustrations, and step-by-step tutorials. We scrape 6,735 pages that interleave text, images, tables, and diagrams. To preserve the layout information, we also save the screenshots of entire pages and extract bounding boxes of the visual elements. There are two files in our Wiki knowledge base. wiki_samples.zip: A sample version of the full knowledge base (10 pages). wiki_full.zip: The full knowledge base (6,735 pages). Check out our paper! @article{fan2022minedojo,
title = {MineDojo: Building Open-Ended Embodied Agents with Internet-Scale Knowledge},
author = {Linxi Fan and Guanzhi Wang and Yunfan Jiang and Ajay Mandlekar and Yuncong Yang and Haoyi Zhu and Andrew Tang and De-An Huang and Yuke Zhu and Anima Anandkumar},
year = {2022},
journal = {arXiv preprint arXiv: Arxiv-2206.08853}
}

项目网站：minedojo.org；论文链接：arxiv.org/abs/2206.08853；GitHub仓库：github.com/MineDojo/MineDojo

《我的世界》（Minecraft）维基百科页面几乎覆盖了该游戏的全部游戏机制，以多模态表格、合成配方、插图及分步教程的形式提供了丰富的非结构化知识资源。我们共爬取了6735个融合文本、图像、表格与示意图的页面。为完整保留页面布局信息，我们同时保存了完整页面的截图，并提取了其中所有视觉元素的边界框。

该维基知识库包含两个文件：
1. wiki_samples.zip：完整知识库的样本版本（含10个页面）
2. wiki_full.zip：完整知识库（含6735个页面）

欢迎查阅我们的研究论文！
@article{fan2022minedojo,
title = {MineDojo：借助互联网规模知识构建开放式具身智能体},
author = {Linxi Fan、Guanzhi Wang、Yunfan Jiang、Ajay Mandlekar、Yuncong Yang、Haoyi Zhu、Andrew Tang、De-An Huang、Yuke Zhu、Anima Anandkumar},
year = {2022},
journal = {arXiv预印本 arXiv:2206.08853}
}

搜集汇总

背景与挑战

背景概述

MineDojo Internet Knowledge Base (Wiki) 是一个基于Minecraft Wiki页面构建的多模态知识库，包含6,735个页面，涵盖游戏机制的文本、图像、表格和图表等信息，并保存了页面布局的视觉元素。数据集提供样本版和完整版两个文件，用于支持开放端智能体研究。