TechTrack

TechTrack数据集由印度理工学院德里分校创建，专注于技术程序中的实体跟踪问题。该数据集包含1351个来自WikiHow的‘计算机和电子’类别的程序，涉及如‘如何连接打印机’等主题。每个程序平均包含4.7个实体，这些实体具有预定义的属性，其值在程序执行过程中发生变化。数据集的创建过程涉及从WikiHow提取程序，并通过众包方式进行实体、属性和属性值的标注。TechTrack数据集的应用领域包括自动化技术支持系统，用于指导用户在执行一系列步骤后应采取的下一步行动，从而提高对程序文本的理解和应用。

The TechTrack dataset, developed by the Indian Institute of Technology Delhi, focuses on the entity tracking task for technical procedures. It contains 1,351 procedures from the 'Computers & Electronics' category of WikiHow, covering topics such as 'How to Connect a Printer'. Each procedure includes an average of 4.7 entities, which possess predefined attributes whose values vary during the procedure's execution. The dataset construction process involves extracting procedures from WikiHow and annotating entities, attributes and their attribute values via crowdsourcing. Application scenarios of the TechTrack dataset cover automated technical support systems, which guide users to determine the next steps to take after completing a series of operational steps, thus improving the comprehension and application of procedural texts.