Peer assessment is a process in which students rate their peers which has many benefits for both the assessor and the assessed. It actively engages students, increases motivation by giving a sense of ownership of the assessment process, encourages autonomy and critical analysis skills, broadens their understanding of the topic, enhances problem-solving and self-assessment abilities as well as develops soft skills. Peer assessment is also beneficial to the teachers as it reduces the strain of the repetitive grading process thus opening more resources for teaching and the development of course materials, especially in large courses with hundreds of students. Sometimes, peer assessment is the only viable option, as in MOOCs (Massive Open Online Courses). Peer assessment has long been studied at all education levels and is gaining traction in recent years, where MOOCs and even the Covid pandemic, which instigated the development of digital competencies, had a positive influence. However, peer assessment is still a demanding process to carry out, but one that can be assisted by modern Internet technology. This paper has a twofold contribution: we present the peer assessment module of our open-source automated programming assessment system Edgar which has been heavily used and developed for the last six years, while peer assessment has been used for the previous three years. Additionally, we present a methodology used and a two-year case study of peer assessment of open-ended assignments in the undergraduate Databases course where 500+ students per season had to provide an entity-relationship model for a given domain and assess their peers' submissions. We discuss our grading methodology, provide in-depth data analysis, and present the students' opinions of the process acquired through an anonymous questionnaire. We find that the process is both demanding in terms of the design of assignments and assessment questionnaires and rewarding in the assessment phase, where the students’ grades turned out to be of high quality.

中文翻译：

---

开放式作业的同行评估方法：使用新型开源系统的大学课程中两年案例研究的见解

同伴评估是学生对同伴进行评分的过程，这对评估者和被评估者都有很多好处。它积极吸引学生，通过给予评估过程的主人翁意识来增加动力，鼓励自主性和批判性分析技能，拓宽他们对主题的理解，增强解决问题和自我评估的能力以及发展软技能。同行评估也对教师有利，因为它减少了重复评分过程的压力，从而为教学和课程材料的开发提供了更多资源，特别是在拥有数百名学生的大型课程中。有时，同行评估是唯一可行的选择，例如 MOOC（大规模开放在线课程）。长期以来，各个教育级别都对同伴评估进行了研究，并且近年来越来越受到关注，慕课甚至是促进数字能力发展的新冠疫情都产生了积极影响。然而，同行评估仍然是一个要求很高的过程，但可以得到现代互联网技术的帮助。本文有双重贡献：我们提出了我们的开源自动化编程评估系统 Edgar 的同行评估模块，该模块在过去六年中已被大量使用和开发，而同行评估已在过去三年中使用。此外，我们还介绍了本科数据库课程中开放式作业的同行评估所使用的方法和为期两年的案例研究，其中每个学期有 500 多名学生必须提供给定领域的实体关系模型并评估他们的同行意见书。我们讨论我们的评分方法，提供深入的数据分析，并通过匿名问卷呈现学生对过程的看法。我们发现这个过程对作业和评估问卷的设计要求很高，而且在评估阶段也很有价值，学生的成绩是高质量的。