jacob-valdez/synthux-economy-r5-w039
收藏资源简介:
SynthUX计算机使用数据集是一个基于合成生成的计算机使用轨迹数据集,专注于经济模拟场景。每个数据记录代表一个工作者在模拟公司内的设备会话:一个目标被扩展为节点树,终端节点通过低级别鼠标/键盘输入驱动真实的桌面模拟器应用程序(包括终端、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和每个节点的帧。数据集支持应用程序原生多应用程序执行(无工作台覆盖)、每个参与者内容多样性、跨操作系统消息传递以及大型多公司经济模拟。每个经济模拟作为独立的数据集存储库发布。数据集内容包括data.jsonl文件(每个渲染轨迹一个记录)、screenshots目录(每个轨迹的每个终端节点帧,在节点的最后一个输入事件处拍摄)和videos目录(完整屏幕录制)。记录模式包括轨迹ID、参与者ID、公司、角色ID、职位、环境(模拟器类型)、自然语言目标、节点树、密集低级别输入事件、因果模拟器状态、对齐链接以及媒体引用。统计显示本版本包含36个轨迹、997个帧、6家公司,环境分布为browser-os(4个)、macos-web-next(17个)和windows-web-next(15个)。数据使用SynthUX生成,观察是因果的(突变通道为低级别输入),捕获过程中未使用高级别应用程序突变。
The SynthUX Computer Usage Dataset is a synthetically generated computer usage trajectory dataset focused on economic simulation scenarios. Each data record represents a device session of a worker within a simulated company: A target is expanded into a node tree, where terminal nodes drive real desktop simulator applications (including terminals, Notes, VS Code, browsers, Slack/Teams, emails, spreadsheets, slides, etc.) via low-level mouse/keyboard inputs. The observed trajectories are recorded as screen videos and frames for each node. The dataset supports native multi-application execution (no workspace overlay), content diversity 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 contents include the data.jsonl file (one record per rendered trajectory), the screenshots directory (frames of each terminal node for each trajectory, captured at the last input event of the node), and the videos directory (full screen recordings). Recorded fields include trajectory ID, participant ID, company, role ID, job title, environment (simulator type), natural language target, node tree, dense low-level input events, causal simulator state, alignment links, and media references. Statistics show that this version contains 36 trajectories, 997 frames, and 6 companies. The environment distribution is browser-os (4), macos-web-next (17), and windows-web-next (15). The data is generated using SynthUX, with observations being causal (mutation channel is low-level inputs), and no high-level application mutations were used during the capture process.




