WONDERBREAD
收藏资源简介:
WONDERBREAD是由斯坦福大学创建的一个大型数据集,包含2928个人类演示的598个未注释的工作流程,旨在评估多模态基础模型在业务流程管理任务中的表现。数据集内容丰富,每个演示包括完整的屏幕录制、操作轨迹和手动编写的标准操作程序(SOP)。创建过程涉及从WebArena基准中筛选工作流程,并通过多轮质量保证确保演示的高质量。该数据集主要应用于业务流程的文档化、知识传递和改进,旨在解决现有ML基准在BPM任务评估中的不足。
WONDERBREAD is a large-scale dataset created by Stanford University. It encompasses 598 unannotated workflows supported by 2928 human demonstrations, and is intended to assess the performance of multimodal foundation models in business process management (BPM) tasks. The dataset is rich in content, with each demonstration comprising full screen recordings, operational traces, and manually written Standard Operating Procedures (SOPs). Its development involved screening workflows from the WebArena benchmark, and implementing multi-round quality assurance to ensure the high quality of the demonstrations. This dataset is primarily utilized for the documentation, knowledge transfer and improvement of business processes, and aims to address the limitations of existing machine learning (ML) benchmarks in evaluating BPM tasks.
WONDERBREAD 数据集概述
WONDERBREAD 是一个用于评估多模态模型在业务流程管理(BPM)任务上的基准和数据集。该数据集旨在支持人工智能在企业应用中增强而非替代人类劳动的目标。
数据集内容
数据集组成
- 2928个人类演示,涵盖598个网页导航工作流程,源自 WebArena。每个演示包含:
- 意图(Intent):工作流程目标的简短自然语言描述。
- 录制(Recording):标注者执行工作流程的全屏录制。
- 动作追踪(Action Trace):所有动作(点击、按键、滚动)和网页状态的日志。
- 关键帧(Key Frames):从录制中提取的每个动作时间戳的图像。
- 标准操作程序(SOP):详细记录标注者所采取步骤的书面指南。
- 演示排名:162个工作流程的演示质量排名。
- 120个问答对:关于工作流程特征的问答对。
数据集分发
demos.zip:完整数据集。gold_demos.zip:仅包含162个工作流程意图的黄金SOP演示。debug_demos.zip:少量演示,用于快速调试。
基准任务
WONDERBREAD 包含以下6个任务,涵盖高层次的BPM应用场景:
- 文档化(Documentation):生成标准操作程序(SOP),以满足质量控制和审计要求。
- 知识转移(Knowledge Transfer):回答关于工作流程操作的用户查询,简化入职流程并减少知识工作者每周等待同事信息的时间。
- 流程改进(Process Improvement):分析工作流程以识别低效和纠正执行错误。
评估
所有评估脚本可在 wonderbread/benchmark/eval 目录中找到。
引用
如果您发现此工作或代码对您有帮助,请考虑引用以下内容:
@article{hazyresearch2024wonderbread, title={Do Multimodal Foundation Models Understand Enterprise Workflows? A Benchmark for Business Process Management Tasks}, author={Michael Wornow and Avanika Narayan and Ben Viggiano and Ishan S. Khare and Tathagat Verma and Tibor Thompson and Miguel Angel Fuentes Hernandez and Sudharsan Sundar and Chloe Trujillo and Krrish Chawla and Rongfei Lu and Justin Shen and Divya Nagaraj and Joshua Martinez and Vardhan Agrawal and Althea Hudson and Nigam H. Shah and Christopher Re}, journal={arXiv preprint arXiv:2406.13264}, url={https://hazyresearch.stanford.edu/wonderbread-website}, year={2024} }
@article{zhou2023webarena, title={WebArena: A Realistic Web Environment for Building Autonomous Agents}, author={Zhou, Shuyan and Xu, Frank F and Zhu, Hao and Zhou, Xuhui and Lo, Robert and Sridhar, Abishek and Cheng, Xianyi and Bisk, Yonatan and Fried, Daniel and Alon, Uri and others}, journal={arXiv preprint arXiv:2307.13854}, year={2023} }




