Helping Improve and Scale Introductory Programming Courses through Automated Code-Reading Exercises

该项目旨在通过开发一种工具来服务于国家利益，该工具将帮助计算机部门为更多的学生进入技术劳动力做好准备，特别是提高他们的编程能力。许多人认为EipE问题通过强调“代码阅读”在帮助学生学习编程方面发挥了重要作用，“代码阅读”是支持“代码编写”的重要发展技能。然而，虽然许多编程课程活动（例如代码编写）可以用一种直接的方式客观地评分，但像“用简单的英语解释”（EipE）这样要求学生阅读给定的代码并用英语描述其功能的问题，很难始终如一地评分。虽然EipE问题受到研究人员的好评，但它们并没有在教学中广泛使用，可能是由于手动评分的负担以及提供给学生的反馈缓慢。该项目的目标是开发一个基于自然语言处理（NLP）的EipE问题自动评分器，为学生提供即时反馈，减轻在课堂上使用EPiE问题的教学负担，并帮助增加这种有效教学方法的使用。从已经足够精确地用于低风险评估的基于单词袋和双字母的实现开始，该项目将使用公开可用的预训练的Transformer体系结构来改进该实现。该项目围绕两个研究问题展开：(1)学生在代码阅读活动中的表现与在编程入门课程中其他活动中的表现有何关系？(2)我们能否通过引入自动化的形成性EipE评估来提高学生在编程入门课程上的成功率？所提出的定性和定量研究将有助于理解新手如何学习编程以及学习阅读代码在学习中所起的作用。NSF IUSE: EHR计划支持研究和开发项目，以提高所有学生STEM教育的有效性。通过参与学生学习轨道，该计划支持有前途的实践和工具的创建，探索和实施。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This project aims to serve the national interest by developing a tool that will help computing departments to prepare more students to enter the technical workforce, in particular to improve their ability to program. Explain in plain English" (EipE) questions are considered by many to play an important role in helping students to learn to program by emphasizing “code reading,” an important developmental skill that supports “code writing.” However, while many programming course activities (e.g., code writing) can be objectively graded in a straight-forward manner, activities like "Explain in plain English" (EipE) questions that ask students to read a given piece of code and describe its function in English are difficult to grade consistently. While EipE questions are well regarded by researchers, they are not in widespread use instructionally, presumably due to the burden of manually grading them and the slow feedback provided to students. The goal of this project is to develop a natural language processing (NLP) based autograder for EipE questions that will provide students with immediate feedback, ease the instructional burden of using EPiE questions in the classroom, and help increase the use of this effective pedagogical approach.Starting from a bag-of-words and bigram-based implementation that is already accurate enough for use in low stakes assessments, this project will refine this implementation using publicly available pre trained Transformer architectures. The project centers around two research questions: (1) How does student performance on code reading activities relate to performance on other activities in introductory programming courses? and (2) Can we improve student success rates in introductory programming courses by introducing automated formative EipE assessments? The proposed qualitative and quantitative studies will contribute to the understanding of how novices learn to program and the role that learning to read code plays in learning. The NSF IUSE: EHR Program supports research and development projects to improve the effectiveness of STEM education for all students. Through the Engaged Student Learning track, the program supports the creation, exploration, and implementation of promising practices and tools.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.