In engineering education, Generative Artificial Intelligence (GenAI) might be misused to complete assessments with limited understanding. On courses that allow the use of GenAI, students might also forget to acknowledge its assistance. There is a need to identify such assistance. We present an automated detector with over-uniqueness and sample matching. GenAI-assisted submissions are identified based on their uniqueness and their similarity to a GenAI sample. Unique to our GenAI detector, it requires no training data and/or dedicated rules for each programming/scripting language. Further, the method can be integrated into any existing similarity detectors to identify plagiarism. The detector covers five similarity measurements, two similarity modes, and eight programming/scripting languages. Our evaluation of four data sets with thousands of submissions shows that our detector is effective (71% MAP). However, many factors can affect its effectiveness, including submission length and student attempts to align the code. Combining both mechanisms does not result in higher effectiveness, yet it takes longer to process.

Feedback is a cornerstone of effective learning, offering students insight into their progress, while also providing instructors with information to refine their teaching. This paper presents a practical tool leveraging large language models (LLMs) to cluster and summarize teaching assistant (TA) feedback. Designed for educators, the tool streamlines the identification of common student issues, provides course-level insights, and generates actionable summaries intended to be modified by educators and shared with students. To validate the tool, we conducted an instructor survey assessing its perceived usefulness and accuracy, compared TA and LLM-generated feedback on a shared assignment, and presented a case study where the tool informed improvements to the assignment handout. Our findings suggest that this practitioner-focused tool can enhance feedback workflows, promote consistency in instruction, and support scalable improvements in teaching and learning.

Generative Artificial Intelligence (GAI) becomes popular due to its ability to assist humans in completing tasks. In some courses, GAI has been employed as part of the education process. However, no studies explicitly focus on promoting GAI literacy to students via their own experiences. We present a method to promote such literacy at the basic level: how GAI can be used responsibly and effectively. Students are encouraged to use GAI in some assessments, and to contrast their experience to assessments without GAI. The method was employed on an English course offered to Information Technology undergradu-ates. Our study found that GAI positively affects student work productivity, and the impact can be more substantial with human validation.

To deal with programming plagiarism, a number of similarity detectors have been developed. However, many of them are either not user-friendly or require local installation. We present a web-based similarity detector that is user-friendly and requires no local installation. The detector is expanded from our previous work, whose performance has been proven. Unlike existing web-based similarity detectors, ours is not commercialized. Further, it offers several similarity measurements and enhanced privacy. According to a controlled experiment with 24 tutors or staff, our similarity detector is effective and efficient. Participants with the detector can identify all copied submissions in which the completion time is reduced by at least two-thirds.

Due to the recent improvements and wide availability of Large Language Models (LLMs), they have posed a serious threat to academic integrity in education. Modern LLM-generated text detectors attempt to combat the problem by offering educators with services to assess whether some text is LLM-generated. In this work, we have collected 124 submissions from computer science students before the creation of ChatGPT. We then generated 40 ChatGPT submissions. We used this data to evaluate eight publicly-available LLM-generated text detectors through the measures of accuracy, false positives, and resilience. Our results find that Copy Leaks is the most accurate LLM-generated text detector, G PTKit is the best LLM-generated text detector to reduce false positives, and GLTR is the most resilient LLM-generated text detector. We note that all LLM-generated text detectors are less accurate with code, other languages (aside from English), and after the use of paraphrasing tools.

To measure the feasibility of AI chatbots in K-12 education, the perspective of both teachers and students regarding the matter should be considered. There are a number of surveys capturing academic perspective of AI chatbots but none of them are focused on K-12 education. Further, they only cover student perspective. We summarized the perspective of K-12 teachers and students via two questionnaire surveys with 75 teachers (44 primary school and 36 high school) and 433 students (242 primary school and 191 high school). K-12 teachers were moderately aware about AI chatbots and somewhat encouraged the use. High school students had similar level of awareness but primary school students were less aware about it. Policies for dealing with AI misuses need to be clearly defined and employed.

To maintain academic integrity, students involved in plagiarism should be identified and penalized. A number of similarity detectors have been developed for that purpose. However, only a few of them are dedicated to C# programming language although the language is often used in courses about application development. Existing C# detectors either are time-inefficient, do not work on incomplete code, or have data privacy concerns. This paper presents a C# similarity detector that can work with large and incomplete submissions offline. Our evaluation shows that the detector is effective in identifying suspected submissions and reporting similar GitHub projects. It is also time efficient as it can process 208 submissions with 9.3 MB C# code in 22 seconds.

Several studies report code quality issues in academia by analysing student submissions. However, most of them focus on novices or a specific integrated development environment (IDE), and the findings might be less representative of code quality issues in general undergraduate computing. This study summarizes code quality issues from seven programming courses with various level of complexity. There are 931 assessment tasks with 15,323 Java/Python program files involved. The reported issues are specifically tailored to computing undergraduates and are selected with checkstyle (Java) and Flake8 (Python). Our study finds that students often neglect to use blank lines between code components, braces where they are optional, and a space after a comment marker. They also sometimes include too much code in one line. This initial study will be expanded via the development of a tool that can automatically summarize the code quality issues of each student submission.

Entering post-pandemic era, some countries consider combining online learning with onsite learning. Given that educators play an important role in the success of such combination, it is important to summarise their perspective about online learning and consider that in designing such combination. This quantitative study summarises perspective of 210 educators about that matter in Indonesia. In general, educators see benefits of online learning except in promoting educators’ teaching presence, supporting vulnerable students, and maintaining student integrity. These issues should be addressed first before combining online learning with onsite learning.

Historically, computing instructors and researchers have developed a wide variety of tools to support teaching and educational research, including exam and code testing suites and data collection solutions. Many are then community or individually maintained. However, these tools often find limited adoption beyond their creators. As a result, it is common for many of the same functionalities to be re-implemented by different instructional groups within the CS Education community. We hypothesize that this is due in part to accessibility, discoverability, and adaptability challenges, among others. Further, instructors often face institutional barriers to deployment, which can include hesitance of institutions to utilize community developed solutions that often lack a centralized authority. This working group will explore what solutions are currently available, what instructors need, and reasons behind the above-mentioned phenomenon. This will be accomplished via a literature review and survey to identify the tools that have been developed by the community; the solutions that are currently available and in use by instructors; what features are needed moving forward for classroom and research use; what support for extensions is needed to support further CS Education research; and what institutional challenges instructors and researchers are currently facing or have faced in the past in developing, deploying or otherwise using community software solutions. Finally, the working group will identify factors that limit adoption of solutions and ways to integrate and improve the accessibility, discoverability, and dissemination of existing community projects, as well as manage and overcome institutional challenges.

To help address programming plagiarism and collusion, students should be informed about acceptable practices and about program similarity, both coincidental and non-coincidental. However, current approaches are usually manual, brief, and delivered well before students are in a situation where they might commit academic misconduct. This article presents an assessment submission system with automated, personalized, and timely formative feedback that can be used in institutions that apply some leniency in early instances of plagiarism and collusion. If a student’s submission shares coincidental or non-coincidental similarity with other submissions, then personalized similarity reports are generated for the involved submissions and the students are expected to explain the similarity and resubmit the work. Otherwise, a report simulating similarities is sent just to the author of the submitted program to enhance their knowledge. Results from two quasi-experiments involving two academic semesters suggest that students with our approach are more aware of programming plagiarism and collusion, including the futility of some program disguises. Further, their submitted programs have lower similarity even at the level of program flow, suggesting that they are less likely to have engaged in programming plagiarism and collusion. Student behavior while using the system is also analyzed based on the statistics of the generated reports and student justifications for the reported similarities.

When checking student programs for plagiarism and collusion, many similarity detectors aim to capture semantic similarity. However, they are not particularly effective for strongly directed assessments, in which the student programs are expected to be semantically similar. A detector focusing on syntactic similarity might be useful, and this paper reports its effectiveness on programming assessment tasks collected from algorithms and data structures courses in one academic semester. Our study shows that syntactic similarity detection is more effective than its semantic counterpart in strongly directed assessments, with some irregular similarity patterns being useful for raising suspicion. We also tested whether take-home assessments have higher similarity than in-class assessments, and confirmed that hypothesis. Consistency of the findings will be further validated on other courses with strongly directed assessments, and a syntactic similarity detector specifically tailored for strongly directed assessments will be proposed.

Covid-19 pandemic greatly affects student daily life. Instead of physically attend classes, they need to meet the lecturer and learn the course material via online meeting platform. The transition somehow introduces some issues like the difficulty of maintaining their focus. This becomes worse for computing students given that the assessments are not limited to standard essays. They include programming and hardware-based assessments which are more difficult to complete at home as students might not have the required software or hardware. This paper reports any issues experienced by 112 computing students in terms of transitioning from offline to online learning. Our study shows that online learning forces the students to allocate more time to study and complete the assessments. Online learning also introduces other issues like higher stress level but still has a few of positive traits like spending less money to physically attend the classes. Many students argue that programming is the most difficult subject to learn in online environment. In response to the issues, some suggestions are provided for computing lecturers.

This research-to-practice work-in-progress paper presents an automated feedback tool that can be used in many teaching environments by integrating it with a web-based assessment submission system. Each time a student submits their work, they will automatically get feedback about aspects of the code quality. Automated feedback tools have been developed to educate students about code quality. However, integrating such a tool into an existing teaching environment can be challenging as these tools can depend on particular working environments, can be separate from the assessment submission system, or can require historical data. Our initial evaluation shows that the tool can be helpful as students do sometimes neglect to satisfy all code quality requirements. However, some false results are expected for spelling correction as student programs are not written in natural language. According to our quasi-experiments, the tool substantially reduces the number of word misspellings in comments due to their substantial frequency of occurrence.

Some code similarity detectors are designed to address academic integrity in early programming courses by recognising subtle variations, in the assumption that the code similarity in these courses is typically higher than that in later courses. Although the assumption is often used, it has no empirical evidence, and might be misleading. This study empirically investigates the assumption by examining the relationship between code similarity and course semester in seven programming courses with a total of 931 distinct assessment tasks. Our study shows that the argument is not necessarily true since in later courses, some assessment tasks require the students to follow a particular structure, to use external libraries, or to implement specific algorithms taught during the course.

Covid-19 changes the way of learning in schools. Students study in a virtual meeting instead of a physical classroom. Given that the transition is suddenly applied, students experience some challenges. A few surveys have been performed to highlight this but, none of them are focused on high school students in a country that is severely hit by the pandemic. This paper surveyed 333 high school students in Indonesia, a country with more than one million cases. The study shows that online learning is perceived to have higher assignment workload and stress level, but the lecture is less clear and has limited teacher-student interaction. Despite the challenges, students do not spend more time in learning. Several possible solutions are provided in response to the challenges like encouraging teachers to record their meetings or introducing gamification.

It is important to educate students about acceptable practices with regard to programming plagiarism and collusion. However, the current approach is quite demanding since it is manual, relying heavily on instructors. The information is delivered briefly, along with other general information, and students may not understand how it applies to their own cases. There is also no warning when students might be about to breach the rules. This doctoral project proposes a system that provides automated, personalised, and timely feedback about programming plagiarism and collusion. If a submission shares undue similarity with other students’ submissions, all involved students will be given similarity feedback, showing their program with similar code fragments highlighted and the similarities explained in natural language, and they are expected to resubmit. Students whose programs do not show clear similarities will be shown a simulation feedback with comparable information. The system is evaluated with some technical measurements and three quasi-experiments.

When comparing student programs to check for evidence of plagiarism or collusion, the goal is to identify code segments that are common to two or more programs. Yet some code segments are common for reasons other than plagiarism or collusion, and so should not be considered. A few code similarity detection tools automatically remove very common segment, but they are prone to false results as no human validation is involved. This paper proposes a semi-automated approach for excluding common segments, where human validation is introduced before excluding the segments. As existing selection techniques are not detachable from their similarity detection tools, we propose a new tool to independently select the segments (C2S2), along with several adjustable selection constraints to keep the number of suggested segments reasonable for manual observation. In order to independently evaluate automated selection techniques, we propose and apply three metrics. The evaluation shows our selection technique to be more effective and efficient than the basis underlying existing selection techniques, and establishes the benefit of each of its selection features.

When student programs are compared for similarity, certain segments of code are always sure to be similar. Some of these segments are boilerplate code -- public static void main String [] args and the like -- and some will be code that was provided to students as part of the assessment specification. The purpose of this working group is to explore what other code is expected to be reasonably common in student assessments, and should therefore be excluded from similarity checking. The answers will clearly vary with programming language, and perhaps with level of assessment item. Working group members will collect assessment submissions from their own or their colleagues' students, and it is hoped that these submissions will together encompass a wide variety of assessment tasks in a wide variety of programming languages. The working group aims to deliver clear guidelines as to what code can reasonably be excluded from automatic code similarity detection in various circumstances. It also aims to deliver a summary of what sort of code lecturers tend to provide for students when setting an assigned task, and why they provide that code.

Conventional practices for recording student attendance in a classroom, such as roll-call and sign-in-sheet, are proven to be inefficient in terms of time and staff-workload. In addition, they are also prone to human error and bogus attendance, which introduce inaccuracy in the recorded data. A number of studies have been conducted to improve the way we record class attendance. However, some of the proposed solutions are costly and impractical. Most solutions also fail to address bogus attendance. This paper presents a low cost solution in recording student attendance. Attendance is recorded on class photographs, students' faces are automatically located using a face detection technique, and students then registered their attendance by simply identifying their face on the records. Mobile applications were developed for both students and lecturers as the primary interfaces to interact with the system.