题目：研究深度学习分类的同伴反馈与在线学习倦怠之间的关系

研究目的：

探究不同类型的同伴反馈如何让学生受益，以及是否有助于缓解在线学习环境下的学习倦怠。

研究问题：

1.与人类对认知内容和情感状态的编码相比，每种深度学习算法在对同伴反馈信息进行分类时的表现如何？

2.学生的在线学习倦怠与使用DL分类的同伴反馈信息之间存在什么关系？

研究方法：

调查问卷；内容分析；实验研究法

研究对象：

一个学校的150名大三本科学生，年龄为19~23岁。所有学生都有教学理论和网络操作的基本知识和至少有一次在线学习经验。

研究过程：

（1）在研究进行之前，将研究目的和研究程序的信息传达给所有学习者，并获得他们的同意。

（2）在在线学习过程中，学生们要观看讲座视频，完成单元测试，提交作业并评价同龄人的工作。鼓励学生在论坛上与其他不同背景的学习者进行强制性讨论。

（3）为了支持高质量的同伴反馈，在评估活动前将同伴反馈指南发送给学生。为学习者提供了每个任务的同伴反馈标准，包括评估指标、要点以及他们给予同伴反馈时的相应分数。

（4）两个同行评审的任务，安排在第5周和第8周，学生可以收到同伴的反馈分数和与他们的作业对应的反馈信息。

（5）在同伴评估后，给学生一份包含学习倦怠和人口统计学（性别、年龄、学习经验等）问题的问卷。

  150名三年级本科生在线课程中同伴反馈的数据作为数据集D2；大学MOOC中两门在线课程的同伴反馈数据集作为D1。利用数据集D1对同伴反馈进行标记，并训练、测试DL分类的有效性；并评估由DL模型自动获得的类别与由人类编码器通过一致性系数获得的类别之间的一致性。

  利用数据集D2，基于认知内容和情感状态的编码规则，验证了预测模型在课程间的可移植性。

研究结果：

（1）通过分析数据集D1，对比了BERT、Bi-LSTM和BERT-Bi-LSTM等深度学习模型在评论情感和认知内容分类任务上的性能。结果表明，BERT在评论情感分类任务中表现最好。在认知内容分类任务中，BERT和BERT-Bi-LSTM在F-度量上表现良好，但BERT在精确度和准确度上优于BERT-Bi-LSTM。在评估DL模型自动分类与人工编码者分类之间的一致性问题上，结果显示BERT模型在情感状态和认知内容分类中取得了最佳的一致性系数。基于以上结果，研究选择使用BERT作为情感状态和认知内容分类任务的预测模型，并认为该模型是可靠的。并在数据集D2中验证了预测模型在不同课程间的可移植性，指出该预测模型在其他课程中有较好的可移植性。

（2）学生在线学习倦怠与收到的同伴反馈信息的关系

通过同伴反馈的认知内容和情感状态可以预测缓解在线学习倦怠。研究结果表明同伴的反馈对于在线学习倦怠的预测和缓解具有重要作用。为了预防和减轻在线学习中学生的倦怠感，重要的是鼓励学生提供包含适当建议的反馈，同时避免在反馈中表达过多正面或负面的个人情绪。此外，避免提供仅仅包含简单整体称赞的强化性或正面反馈是有益的。

结论：

在本研究采用深度学习模型和多元回归分析来检验同伴反馈与在线学习倦怠之间的关系。首先，训练后的模型在准确分类同伴反馈方面表现出了良好的性能。其次，MLR结果显示，同伴反馈的认知和情感特征与在线学习倦怠存在多重关系。具体来说，提供暗示性的反馈和避免过度的积极或消极的反馈有助于减少学习倦怠。这些发现强调了在设计在线同伴评估活动时，同时考虑同伴反馈的认知和情感方面，以及在在线学习环境中实施预防和减轻学生职业倦怠的教学策略的重要性。

不足之处：

（1）使用的数据样本是有限的。选择在某一特定大学参加同一在线课程的学生，这可能不利于研究结果的泛化。此外，同伴反馈信息的分布存在不平衡，教学反馈和请求反馈很少。

（2）本研究没有对对照组进行准实验设计。作者认为，在这项研究中，同伴反馈是在线课程学习的一个重要组成部分，而让没有接受同伴反馈的对照组参与进来是一个挑战。

（3）本研究主要关注接收反馈的分类及其在缓解自我报告的学习倦怠中的作用，而没有考虑学生在在线环境中的社会互动过程。

（4）仅仅依靠调查来衡量在线学习倦怠，可能会导致缺乏客观性。

Title:The role of educational robotics in cultivating K12 students' computational thinking : A systematic literature review

Keywords: computational thinking ; educational robotics ; Literature review ; Robotics in education

1. Introduction

The world has undergone profound transformations due to the advent of digital technology. The widespread availability of robotics, artificial intelligence, and mobile communication technology has created an environment where in children are growing up surrounded by smart technology, the Internet, and big data. This digital technology has significant implications for their childhood experiences, and its continued advancement will undoubtedly impact their cognitive abilities and future development. As digital technology assumes an increasingly prominent role in society, CT has emerged as a crucial competency that individuals in this era must possess, extending beyond the purview of computer scientists. It is a determining factor in one's adaptability to the future landscape. Consequently, the field of education is placing escalating emphasis on the effective cultivation of CT skills among K-12 students.

1.1 Computational thinking

In the century of digital advancements, CT has emerged as a pivotal future capability, garnering consensus among the international community. It not only fosters effective problem-solving but also nurtures the development of creative thinking, logical reasoning, and analytical skills. As a versatile cognitive approach, CT finds applications across diverse disciplines and industries. The ability to tackle problems, engage in critical analysis, communicate effectively, and think creatively has become indispensable in today's competitive and rapidly evolving digital landscape. Consequently, numerous countries have integrated CT training into their compulsory education curriculum.

In 1980, Seymour Papert first introduced the concept of CT. He devised "Logo," a programming language specifically tailored for children. Papert proposed combining computers with children's learning experiences to cultivate their creativity, critical thinking, and transform their knowledge acquisition approach through "Logo" . According to Papert, children can engage in geometric experimentation and enhance their logical thinking skills while writing programs to interact with the virtual "turtles." Therefore, he emphasized the cultivation of CT in programming among students.

In 2006, Wing  put forth a proposition in the renowned publication Communications of the ACM, asserting that "CT is a systematic approach that leverages fundamental principles and methodologies derived from the field of computer science to solve complex problems, design intricate systems, and gain insights into human behavior." The key of this cognitive paradigm lies in the processes of abstraction and automation. The introduction of this concept elicited widespread astonishment due to its exceptional nature. Building upon this seminal work, Wing  later refined the concept in 2010, positing that "CT refers to a cognitive process of formulating problems and devising solutions in a manner that renders them amenable to effective execution by information processing agents." Consequently, computers have transcended their role as mere tools for human assistance, and the underlying CT inherent in them has transformed into a form of analytical thinking. Wing fervently encourages individuals from all walks of life to cultivate and harness the prowess of CT. These foundational perspectives serve as the theoretical bedrock of CT and provide a firm basis for subsequent explorations in CT's pedagogical advancements, specifically focusing on optimizing student training within particular disciplines.

1.2 Robotics Education

With the further development of artificial intelligence technology and human-computer interaction technology, the field of robotics has ushered in further development opportunities. As a branch of robotics in the field of education, ER has become a teaching tool that enriches teaching forms and provides learners with understanding of technology and practical experience. The robots currently used in classroom teaching are classified according to their functions, mainly including programming robots, interactive robots, and auxiliary robots. Classified according to appearance, it mainly includes bionic robots and social robots. In recent years, a considerable amount of research has focused on introducing educational robots into classrooms to assist teaching, in order to understand what benefits can be obtained from robot activities. As a scaffolding strategy, ER can motivate students of all ages to learn, limit their attention, and develop their CT and teamwork abilities.

Previous research has shown that with the increasing focus on STEM (Science, Technology, Engineering, Mathematics) education, robotics is considered an innovative solution. Robot education can provide learners with practical experience in understanding technology and mechanical language, and proactively build knowledge in practice; adapt to the constant changes brought about by complex environments; and utilize knowledge in real situations or across time, space and environments. In other words, robot education can integrate robotics technology with the knowledge and skills of other subjects, attracting students with different interests and learning styles. This effect even has the same effect on young learners .

This view was quickly confirmed. In recent years, many researchers have focused on young students to explore the role of robotics technology in integrating different disciplines in cultivating key abilities of young learners, and the experimental results are quite impressive.In ER integrated programming education, research has found that compared with screen-based programming, using physical ER as the output object of programming can better promote communication among peers . When students use Lego kits to build robots and use Lego software to code and solve problems, not only do they develop their critical thinking and computational thinking, but their communication and collaboration skills and teamwork skills are also improved . In addition, in the context of ER integrated pair programming, students' CT skills and target language abilities are promoted through classroom activities using interdisciplinary board games. And in the context of pairing, where two people communicate and help each other, it also helps young learners reduce their foreign language learning anxiety.

2. Methods

Since the topic of this research is relatively clear, we searched in multiple databases such as Web of science, Sage, and Science Direct. When searching, there is no limit on the publication time of the documents, and the document types are Article. Review Article, Dissertation Thesis, Early Access and Correction were not included. We mainly search for two topics:(a) terms describing computational thinking(b) terms describing robotics (i.e., educational robot*, robot*, robot*-based learning, learning in robot*, educational use of robot*, integrating the robot* and education), and use them with the Boolean directive "AND" "connect. A total of The criteria for selecting literature are as follows: (1) The type of literature is empirical research, and the research methods include experimental method, quasi-experimental method and case study method. (2) The learner’s academic stage is K12. (3) The robot used in the experiment is a physical robot, not a virtual robot.

3. Findings and implications

3.1  文献分析：对每篇文献进行分析和评论，概括其研究方法、主要发现、假设、局限性等内容。

3.2  研究进展和不足：总结当前研究领域的主要研究进展，指出已有研究的不足之处和未解决的问题。

4. Discussion

5. Conclusion

6. Limitations

总结以上研究中的主要发现和结论，说明已有研究的局限性和对未来研究的启示，并提出自己进一步研究的建议。

7. References

题目：

通过教育机器人培育计算思维：创造性计算问题的模型

研究目的：

得到一个模型，使教师能够有效地设计ER活动来促进学生CT的发展，预测活动期间课堂上可能发生的情况，并相应地计划具体干预措施。

研究问题：

根据提议的框架设计并在测试组的实验条件下实施的教学干预措施的有效性。
1.ER 活动的非指导性方法（即无限制地访问编程接口）是否会促进试错方法？
2.编程接口的阻塞是否会促进与问题理解、想法生成和解决方案制定相关的认知过程？
3.部分阻塞（即使用编程接口而不在机器人上执行代码的可能性）是否有助于学生在解决问题的过程中逐步前进？

研究对象：
共有 29 名小学生（13 名女孩和 16 名男孩，年龄在 9 至 10 岁之间）。
在研究之前，所有学生都已通过几门学校课程（每周 1 小时，持续 12 周）了解了 Thymio 机器人和 VPL 编程接口。

研究方法：
准实验、内容分析法

研究过程：

两组小学生使用教育机器人Thymio参加了教育机器人的教学活动，所有学生被随机分配到两人一组或三人一组。一组在没有任何附加约束的情况下完成了任务，另一组则接受了基于所提出的模型开发的教学干预。

对照组 ：
学生有 40 分钟的时间来实践割草机机器人。在整个时间段内，他们可以使用提供给他们的一切：场所、Thymio 机器人和 VPL 编程接口，没有施加额外的限制。
测试组：
前 10 分钟进行编程接口阻塞：学生们可以使用机器人的活动场地和 Thymio 机器人，但不允许他们使用 VPL 编程平台。
下一个10分钟：解除封锁，学生可以使用所有东西。
接下来是 10 分钟的部分封锁阶段，学生可以访问包括 VPL 平台在内的所有内容，但不允许他们在机器人上执行任何代码。
最后10分钟，封锁再次解除，学生们可以使用一切设施。

每个小组以及他们与 VPL 平台和操场的互动都被录制成视频以供以后分析。

由两位评估人员分析了各种预先记录的 ER 活动。
两名评估员独立进行行为分析，将学生在机器人割草机活动期间的行为依次映射到 CCPS 模型的不同阶段。

研究结果：
(i)教育机器人活动的非教学方法（即无限访问编程接口）促进了试错行为
（ii）编程接口的预定阻塞促进了与问题理解、思想生成和解决方案制定相关的认知过程
（iii）逐步调整编程接口的阻塞可以帮助学生建立一个妥善解决的策略来解决教育机器人问题，并可能是提供脚手架的有效方法

结论：
本研究的结果提供了关于对 ER 活动进行具体教学干预的必要性的初步证据，说明教师如何使用所提出的模型来设计旨在发展 CT 技能的 ER 活动。因此可以从理论框架过渡到更具操作性的框架。

不足之处：
1.实验的样本较小。
2.教师是否真正意识到该模型对其教学活动的附加价值有待商榷。