Peer code review, which involves programmers examining and evaluating one other's code for quality, has been adopted in Computer Science (CS) classrooms to provide authentic learning experiences, support collaboration, develop programming skills, and facilitate peer learning. Although a body of research has explored the effectiveness of peer code review as learning activities, limited work focused on how the role of reviewer/reviewee impacts students' learning during the peer code review process. This work shares preliminary findings from the implementation of peer code review in a senior CS course. The instructor designed a double-blind peer code review activity in which students first individually review an assigned submission by completing a code review checklist and then, engage in group discussion. Through surveys, we examined students' perceptions of learning through peer code review. We found that students highly valued the opportunity to review others' solutions, thought it somewhat challenging to provide quality feedback, and had mixed opinions about being reviewed.

Sequential recommendation systems play a key role in modern information retrieval. However, existing intent-related work fails to adequately capture long-term dependencies in user behavior, i.e., the influence of early user behavior on current behavior, and also fails to effectively utilize item relevance. To this end, we propose a novel sequential recommendation framework to overcome the above limitations, called ICMA. Specifically, we combine temporal variability with position encoding that has extrapolation properties to encode sequences, thereby expanding the model’s view of user behavior and capturing long-term user dependencies more effectively. Additionally, we design a multi-view data augmentation method, i.e., based on random data augmentation methods (e.g., crop, mask, and reorder), and further introduce insertion and substitution operations to augment the sequence data from different views by utilizing item relevance. Within this framework, clustering is performed to learn intent distributions, and these learned intents are integrated into the sequential recommendation model via contrastive SSL, which maximizes consistency between sequence views and their corresponding intents. The training process alternates between the Expectation (E) step and the Maximization (M) step. Experiments on three real datasets show that our approach improves by 0.8% to 14.7% compared to most baselines.

Live coding has gained its prominence in Computer Science (CS) classrooms as it enhances learning experiences by providing real-time demonstrations of programming and debugging during lectures. However, live coding may also present challenges to the effectiveness and inclusiveness of the classroom environment. This experience report presents the use of live coding in an introductory level CS course to better understand its impact on novice learners and their perceptions of the strategy. We conducted a between-subjects study at a public university in the United States across two course offerings. Specifically, the instructor taught the same course in two consecutive quarters: one with live coding and one without. Through assignment and exam scores, grades, and surveys, we compared data from students who experienced live coding with those who did not. We found that while live coding prompted engagement and interaction, and students believed that it helped them learn better, there was no statistically significant difference in course performance in terms of assignment and exam scores, nor in final grades. Additionally, most students prefer to keep live coding in lectures despite the challenges it presents. We also shared challenges, lessons, and practical instructional strategies learned from this experience, in the hope that they will contribute to developing more engaging and inclusive learning experiences in CS classrooms.

Despite the similar global structures in Chest X-ray (CXR) images, the same anatomy exhibits varying appearances across images, including differences in local textures, shapes, colors, etc. Learning consistent representations for anatomical semantics through these diverse appearances poses a great challenge for self-supervised pre-training in CXR images. To address this challenge, we propose two new pre-training tasks: inner-image anatomy localization (IIAL) and cross-image anatomy localization (CIAL). Leveraging the relatively stable positions of identical anatomy across images, we utilize position information directly as supervision to learn consistent semantic representations. Specifically, IIAL adopts a coarse-to-fine heatmap localization approach to correlate anatomical semantics with positions, while CIAL leverages feature affine alignment and heatmap localization to establish a correspondence between identical anatomical semantics across varying images, despite their appearance diversity. Furthermore, we introduce a unified end-to-end pre-training framework, anatomy-aware representation learning (AARL), integrating IIAL, CIAL, and a pixel restoration task. The advantages of AARL are: 1) preserving the appearance diversity and 2) training in a simple end-to-end way avoiding complicated preprocessing. Extensive experiments on six downstream tasks, including classification and segmentation tasks in various application scenarios, demonstrate that our AARL: 1) has more powerful representation and transferring ability; 2) is annotation-efficient, reducing the demand for labeled data and 3) improves the sensitivity to detecting various pathological and anatomical patterns.

This research category full paper explores the application of learning analytics in engineering and computing education over the past decade. Although these fields have gained increasing attention in nowadays technologically advanced environment, research about how learning analytics has been utilized within these disciplines remains limited. To address this issue, we conducted a bibliometric analysis with topic modeling on 903 articles published from 2011 to 2023 at the Learning Analytics and Knowledge conference. Specifically, our analysis identified prevalent research topics, methodological innovations, and pedagogical practices of the use of learning analytics within engineering and computing education. The study provides significant insights into how learning analytics has advanced engineering and computing education and highlight future research directions.

This research category full paper explores the dynamic associations between learning engagement and learning outcomes in an asynchronous computing course. Learning engagement acts as an essential mediator between learning contexts and learning outcomes, providing significant insights into instructional strategies and student success. In this study, we first conceptualized and quantified student learning engagement with a multifaceted framework from behavioral, academic, cognitive, and affective aspects. Then, we examined the temporal dynamics of students' learning engagement using the state space model, and further evaluated its associations with students' weekly learning outcomes. The analytical results of this research provided significant implications for supporting learning and teaching in a large-scale asynchronous computing course.