Supporting Learning of Programming with Tutoring Systems

在许多机构，编程课程现在被认可为所有学生（不仅仅是计算机科学专业的学生）在学前和高等教育水平。然而，编程课程的失败率和退学率是出了名的高。这可能是因为编程虽然与其他解决问题的活动有一些重叠，但为了掌握编程所需的技能，还需要一种新颖的思考和解决问题的方式。由于这些技能是编程所独有的，因此支持它们需要创新的教学方法。我的研究计划将推进教育技术，称为辅导系统，通过鼓励建设性的例子处理来帮助学生学习编程。例子在认知技能习得中发挥着关键作用，但迄今为止，利用它们来支持通过辅导系统学习编程的工作相对较少。为了解决这一差距，建议的工作有三个主要的研究流：(1)专注于示例设计的干预（包括示例解决方案中的差距，鼓励自我解释的提示和错误示例），(2)示例与问题的教学顺序，以及(3)通过多个示例促进编程作为一种创造性活动。在短期内，重点将放在辅导系统功能的设计上，包括研究识别界面特征，以促进学生通过建设性推理学习。长期目标是增加辅导系统的适应能力，通过扩展导师与学生模型来实现定制教学。学生模型存储有关单个学生的信息（例如，他们的知识和影响），并基于学生与辅导系统的交互实时更新该信息；辅导系统使用这些信息来根据特定学生的需求定制教学。这项拟议的研究具有智力上的价值。我的研究项目开发的辅导系统将作为回答与有效辅导系统设计相关的研究问题的试验台，以及专注于核心编程技能的新型认知和情感干预的效用。因此，研究结果将有助于人工智能与教育、认知科学和人机交互领域。在广泛的影响方面，我的研究项目将产生一系列针对不同受众的在线导师，为所有学生提供个性化的帮助，包括来自非传统CS背景的学生。通过使编程易于使用，这有可能激发对这一领域的兴趣，从而扩大对该领域的参与。

In many institutions, programming classes are now endorsed for all students (not just computer science majors) at pre- and post-secondary levels. However, programming classes have notoriously high failure and drop-out rates. This may be because programming, while sharing some overlap with other problem-solving activities, additionally requires a novel way to think and problem solve in order to master the skills needed to program. Since these skills are unique to programming, supporting them requires innovative pedagogical approaches. My research program will advance educational technologies, called tutoring systems, that help students learn to program by encouraging constructive processing of examples. Examples play a key role in cognitive skill acquisition but to date relatively little work exists on leveraging them to support learning of programming via tutoring systems. To address this gap, the proposed work has three main research streams: (1) interventions focused on example design (including gaps in example solutions, prompts to encourage self-explanation, and erroneous examples), (2) instructional sequencing of examples vs. problems, and (3) promotion of programming as a creative activity through multiple examples. In the short term, the focus will be on the design of tutoring system functionalities, including studies to identify interface features that promote student learning through constructive reasoning. The long-term goal is to add to the tutoring systems adaptation capabilities that tailor instruction, accomplished by extending the tutors with student models. A student model stores information about individual students (e.g., their knowledge and affect) and updates that information in real time based on students' interactions with a tutoring system; this information is used by the tutoring system to tailor instruction to a given student's needs. The proposed research has intellectual merit. The tutoring systems developed by my research program will serve as testbeds for answering research questions related to effective tutoring-system design, as well the utility of novel cognitive and affective interventions focused on core programming skills. Thus, the results will contribute to artificial intelligence & education, cognitive science, and human-computer-interaction fields. On the broad impact side, my research program will produce a series of online tutors designed for a diverse audience that provide personalized help to all students, including ones coming from non-traditional CS backgrounds. By making programming accessible, this has potential to motivate interest in this domain and thus broaden participation in the field.