一句话：扩展 ScratchLog，结合 Whisker 测试与代码嵌入模型，实时可视化学习者在目标导向练习中的编程进度。
Scratch 是流行的积木式编程入门环境，但不提供学习分析，教师难以掌握学生进度。已有 ScratchLog 主要支持课堂管理和面向开放式任务的一般分析，无法反映目标导向练习的完成情况。本文提出 ScratchLog+ 扩展：利用 Whisker 测试套件收集学生程序功能正确性信息，并使用代码嵌入模型估计其程序相对示例解答在结构和语义上的进度；为便于教师理解，这些通常抽象的嵌入向量被转化为可解释的可视化并直接集成在 ScratchLog 用户界面中。作者提供 screencast 演示该工具，但摘要未报告大规模用户研究或定量评估结果。
While the block-based programming language Scratch is a popular introductory programming environment, gathering information about student progress is challenging since the environment does not expose any kind of analytics. The ScratchLog tool already alleviates this somewhat, but focusses mainly on classroom management and general analytics suited for open-ended tasks. To also provide insights about student progress in goal-oriented exercises, we propose the ScratchLog+ extension in this paper. Our tool combines Whisker test suites to collect information about the functional correctness of students’ programs, and code embedding models to estimate structural and semantic progress towards an example solution. To make these new analytics of ScratchLog+ accessible to teachers, they are directly integrated into the ScratchLog user interface as interpretable visualisations of the typically abstract code embedding vectors. We provide a screencast demonstrating the tool at https://youtu.be/lQSCDqvbpQY.