jacob-valdez/synthux-economy-r4-w047
收藏资源简介:
SynthUX计算机使用数据集是一个基于视觉的计算机使用轨迹数据集,由SynthUX生成。每个记录代表一个模拟公司内部的工作者设备会话:一个目标扩展为节点树,终端节点通过低级鼠标/键盘输入驱动真实的桌面模拟器应用(如终端、笔记、VS Code、浏览器、Slack/Teams、邮件、表格、幻灯片等),观察到的轨迹被记录为屏幕视频以及每个节点的帧。该数据集支持应用原生多应用程序执行(无工作台覆盖)、每个参与者内容的多样性、跨操作系统消息传递和大型多公司经济模拟。每个经济模拟都作为独立的数据集仓库发布。数据集包含data.jsonl文件(每个渲染轨迹一条记录)、截图文件夹(每个轨迹的每个终端节点帧,在节点的最后一个输入事件时拍摄)和视频文件夹(完整的屏幕录制)。记录模式包括轨迹ID、参与者信息、环境、目标、节点树、输入事件、观察、对齐和媒体文件。此构建版本包含36个轨迹、36个视频、932个帧和6个公司,环境分布为browser-os(14)、macos-web-next(11)和windows-web-next(11)。数据来源为SynthUX生成,观察具有因果性(突变通道为低级输入),捕获过程中未使用高级应用程序突变。
The SynthUX Computer Usage Dataset is a vision-based computer interaction trajectory dataset generated by SynthUX. Each record represents a simulated worker’s device session within a fictional company: A target task is expanded into a node tree, where terminal nodes drive real desktop emulator applications (e.g., Terminal, Notes, VS Code, Browser, Slack/Teams, Email, Spreadsheets, Slides, etc.) via low-level mouse and keyboard inputs. The observed interaction trajectories are recorded as screen videos and frame images for each node. This dataset supports native multi-application execution (no workspace overlay), diversity of content per participant, cross-operating-system messaging, and large-scale multi-company economic simulations. Each economic simulation is released as an independent dataset repository. The dataset includes a data.jsonl file (one record per rendered trajectory), a screenshots directory (frame images for every terminal node of each trajectory, captured at the final input event of the respective node), and a videos directory (full screen recordings). Record schemas include trajectory ID, participant information, environment, target task, node tree, input events, observations, alignments, and media files. This release build contains 36 trajectories, 36 videos, 932 frame images, and 6 companies, with environment distributions of browser-os (14), macos-web-next (11), and windows-web-next (11). The dataset is generated via SynthUX, where the observations are causal (the mutation channel consists of low-level inputs), and no high-level application mutations were utilized during the data capture process.




