Programming computational models is a key practice that supports both scientific and computational learning. Block-based, domain-specific programming environments have made this practice more accessible but may constrain what students can design compared to general- purpose or text-based languages. This poster reports on the classroom use of block unpacking-a feature that allows high-level primitives to be opened, inspected, and modified-within a domain-specific modeling environment. We implemented unpacking in a four-day high school biology unit on eutrophication, where students programmed models to represent interactions among algae, bacteria, fish, nutrients, and oxygen. From the 16 students who participated, we analyzed 36 unpacking modifications made by five focal students, classifying each as a parameter change or a code structure change, and examining whether these modifications were informed by scientific reasoning. Our findings show that unpacking let students move beyond fixed domain-specific primitives, producing diverse and more expressive models. Some introduced new mechanisms, such as photosynthesis, while others focused on optimizing or debugging simulations. Importantly, more complex edits did not always correspond to deeper scientific reasoning. This study highlights unpacking as a mid-level design strategy that can expand student agency and flexibility in block-based modeling. We discuss implications for designing programming environments and instructional supports that balance ease of entry with opportunities to engage meaningfully in computing practices.
This poster presents a practical model for transatlantic collaboration in computer science education through a blended mobility program between Karlsruhe University of Education (Germany) and Northeastern Illinois University (USA). The program funded by the European Union (Erasmus+) combines virtual and physical mobility phases, enabling students to participate in collaborative courses such as Physical Computing and Cybersecurity without interrupting their home semester. We share insights from the program's first implementation cycle, including student feedback and preliminary data on learning outcomes. This model demonstrates how EU-US cooperation can effectively support practical, handson computer science courses in a blended learning environment. Our work highlights challenges and best practices for designing sustainable, international blended mobility programs in computer science teacher education, offering valuable implications for institutions seeking to foster global collaboration in computer science education.
College instructors often allow students some form of cheatsheet during an exam, such as one page of notes and formulas. Many believe these reduce test anxiety and help students to do better on exams. Focusing on the latter, we found, in a study of two different types of computer science courses, that features we might have believed would help students, do not actually have any correlation with performance. Coupled with mixed results from the literature, we are still left with the agonizing question: what about cheatsheets are useful?
The rapid expansion of K-12 computer science (CS) education has outpaced the supply of teachers with formal CS training, leaving many educators to teach outside their expertise. This creates challenges in assessing student understanding and interpreting conceptual growth through code. To address this, we introduce ARCHIE (Assessment and Recognition of Coding Habits for Instructional Enhancement), a system that extracts conceptual patterns from student programs and visualizes them at both individual and class levels. We solicited feedback from 20 K-12 CS teachers regarding their confidence in analyzing code and their perceptions of ARCHIE. Teachers reported that ARCHIE helped make student thinking more visible, supported formative assessment, and showed strong potential to enhance instructional decision-making. These findings suggest ARCHIE could provide valuable support for K-12 CS teachers, particularly those without a computing background.
Computing programs increasingly need to prepare graduates who can apply AI in context, not just write software. We present a transferable model for an undergraduate B.S. in Applied AI that treats application domains (e.g., Business, Life Sciences) as co-equal with a rigorous AI technical core. To widen participation while maintaining depth, we replace the traditional calculus sequence with a tailored "Mathematics for AI" track. The 120-credit curriculum is anchored by a technical AI core, required co-operative education, and a culminating one-semester senior design project, with ethics explicitly threaded throughout. For the SIGCSE community, our contribution is a pragmatic blueprint for institutions that aim to broaden participation and produce integrated AI practitioners. We share our domain-based curriculum, course maps, learning outcomes, and launch domains co-developed with partner schools, with institutional adoption planned for Fall 2026.
Team-based projects are essential in computing education, offering students valuable collaboration practice. However, assessing individual contributions within teams remains challenging, especially in remote or large courses. Traditional peer evaluations are widely used but are limited by bias and lack evidence. To address this, we present RIPEL, a data-augmented peer evaluation system that supports evidence-based feedback of teamwork by visualizing log data from digital collaboration platforms. The data captures different components of teamwork, including technical contributions, content generation, and communication. Students review and annotate an interactive timeline of team contributions, grounding their feedback in evidence. A preliminary evaluation suggests that RIPEL can support more meaningful, evidence-based peer evaluations.
Equipping users with the ability to recognize AI hallucinations is essential for AI literacy. Building on research highlighting a lack of studies involving younger learners in the chatbot design process, we explore a constructionist approach to AI literacy focusing on understanding of hallucinations. We engaged 48 middle school learners in designing LLM-based chatbots with customized characters, roles, and constraints using a learning environment, LUMI. Through peer testing and collaborative evaluation, learners developed practical strategies for identifying hallucinations and understanding AI limitations. Pre- and post-surveys revealed substantial improvements in learners' AI understanding, awareness of AI hallucinations, and confidence in designing trustworthy chatbots. Qualitative analysis showed that learners developed foundational understanding of chatbot functionality, LLMs, and prompt design, while demonstrating recognition of data limitations and strategies to mitigate hallucinations. This work contributes to teaching responsible AI interaction for youth learners.
Modern medical devices integrate hardware and software, thereby expanding the attack surface and creating patient safety risks. Regulators set cybersecurity expectations, and international standards guide implementation. However, outcome-based, method-agnostic policies give manufacturers wide latitude, producing variability, ambiguity, and fragmented, non-reproducible practices in training and education. We address this gap with a reproducible, open-source reference platform that operationalizes security-by-design in alignment with current regulatory expectations and standards. We also report an initial pilot course using the platform as a foundation for consistent medical-device cybersecurity curricula.
Frequent practice with small exercises in introductory programming can reinforce classroom concepts, build confidence and independence, and reveal misunderstandings. However, it is often difficult to develop and deploy interesting problems, and utilizing frameworks such as online textbooks can require fees or subscription services that create equity and access barriers to low-income students. The Ursinus WebIDE provides an open-source open educational resource (OER) that supports the rapid development of student exercises in a variety of course contexts and languages, including Java, Python, C++, Scheme, Prolog, and graphics (GLSL, WebGL) and multimedia (Pyodide). These exercises are presented to students in a consistent and serverless browser-based IDE interface that presents scaffolded files in a tree-based filesystem that can be specified by an instructor via a single easy-to-use Markdown file. Student submissions are checked against sample correct and incorrect solutions, with feedback provided for each. Correct submissions are automatically encrypted and immediately uploaded to a Google spreadsheet in a FERPA-compliant way, and processed with our open-source tool to automatically grade submissions to a course learning management system (LMS), without requiring integration with institutional computing systems. Students can be given hints for common mistakes, and they can also be given optional partial credit for attempting a solution. These tools can be deployed entirely free of cost by instructors and students using static Github Pages and Google Forms. We provide a repository of drop-in exercises compatible with the system via a GitHub repository for use, adaptation, and contribution.
We present a Large Language Model (LLM)-assisted SQL learning system that closes the loop from problem discovery to grading. Grounded in real-world data-wrangling scenarios, our agentic workflow (i) synthesizes industry-style practice problems with pedagogical metadata, (ii) produces executable reference SQL via a multi-step operator-planning pipeline, and (iii) grades student submissions against rich rubrics while explaining partial credit and surfacing actionable feedback for revision. We evaluate two core capabilities. First, on a large corpus of realistic SQL problems, our zero-shot, multi-step reference-answer generator, implemented with OpenAI's o4-mini, substantially outperforms a single-prompt baseline while approaching the state-of-the-art pipelines trained with supervised learning. Second, in a classroom deployment, we compare LLM-assisted grading with human graders across four exam questions, encompassing 326 submissions evaluated by six graders. The results indicate that LLMs can provide grading signals competitive with those of human graders for many question types. Overall, the system is designed for responsible educational use through real-world problems, generated reference solutions, and grading assistance. Together, these features enable scalable practice generation and grading, which improves student learning while augmenting instructor capacity.
Teaching computing at scale necessitates some level of automation. Many undergraduate courses have scaled up through commercial Learning Management Systems or, sometimes, "homegrown" tools developed by teaching staff. The development of homegrown tools within universities, especially those that lie in computing classrooms, has been reported on for some time. However, research focuses strongly on the applications of these tools, rather than the design process and the overarching ecosystem of software development in which these tools are embedded. We investigate the design of homegrown tools within a large R1 institution, where much of the development is conducted by and for Teaching Assistants (TAs). Our results are synthesized into a four phase design cycle (Figure 1)-a framework that allows us to identify gaps in development and grounds future exploration of tool development within computer science education.
In a recent pilot study, we investigated computer science (CS) motivation among middle school students through multiple lenses: identity, interest, value, communal/agentic goal affordances and endorsements, sense of belonging, self-efficacy, and intentions to pursue CS. Our research question was: What are the motivational differences to learn computer science among boys and girls in middle school? We collected data from three U.S. sites (California, Michigan, and Illinois) in Spring 2025. We conducted an independent-samples t-tests to ascertain gender differences in each type of CS motivation. While significant results were found across three measures (value for CS, communal goal endorsements, and intentions to pursue CS), the remaining measures did not produce significant differences, indicating that boys and girls may have similar CS identities, interest in CS, sense of belonging in CS, and self-efficacy.
We have developed a novel Automated Assessment Tool (AAT) facilitating formative assessments for novice programming students in a Higher Education (HE) CS setting. This AAT forms an analogue of traditional in-person HE CS computer lab sessions. Traditionally teaching staff would engage with students in-person offering feedback to, and receiving feedback from, students about learning materials. Meaningful two-way communication can be difficult to achieve with individuals in larger HE CS cohorts, which may be taught in a blended way. Our novel AAT offers automatically generated feedback to students and captures feedback from students (per question and per assessment) to inform teaching practice. The novel AAT has been used in four semesters with student volunteers from the authors HE CS setting. A mixed-methods approach has captured qualitative and quantitative data. Key findings are: 1. Parsons problem type questions were associated with no negative feedback from students in initial or repeat experiments, unique amongst the four AAT question-types available. 2. Students report concerns about providing free text feedback per question, but we observe this is a valuable source of information for teaching staff - if automatically collated for presentation in conjunction with other types of feedback from students and learning analytics. This research informs on best practices supporting communication between novice programming students and teaching staff where an AAT is a pedagogical necessity, promoting consideration for automatically collated student feedback, as part of AATs. Future work should replicate the experiments conducted in other HE CS settings.
Theoretical Computer Science educators often need to generate solutions to problems in a variety of forms, requiring diagrams and detailed explanations for every step. Currently, solution generation is largely a manual and time-consuming process on the part of the educator. We introduce AUTO-X, an automatic solution generator and explainer, designed for a variety of standard theory problems and conversions. AUTO-X distinguishes itself by producing not only the visual and formal components of a CS theory solution but also by providing step-by-step explanations of the underlying process. Additionally, it is an easily extensible tool for educators when either adding a new CS theory conversion or algorithm, or adapting our existing templates for institutional-specific terminology/standards.
This panel is inspired by this year ' s SIGCSE readings, especially Whistleblower by Susan Fowler and the Modern Figures podcasts. Broad perspectives from history and current events will drive a conversation about persistence, resistance, empowerment, and maintaining principles. Our panelists will tell stories about individuals and groups who have faced social and financial barriers, persisted, and through resistance changed the world for the better. The goal is to motivate and inspire actions that improve CS education for all.
Service-learning effectively attracts and retains students in computing, thereby contributing to broadening participation in computing (BPC). This work presents a sustainable, project-based outreach model designed to engage students in educating communities about computing concepts. The model is grounded in a successful, multicomponent service-learning program established at our undergraduate college to address the need to increase the number of students who persist in information technology (IT) and related majors. Our unique program has proven sustainable through an established elective course and strong departmental support. This work reviews the foundational components that have supported the outreach model over its 13 years of existence. More significantly, we highlight its recent enhancements, including a new framework that unifies the format of our outreach activities and an online platform to document existing and new initiatives, facilitating adoption by others. The newly introduced outreach components include a series of career development events featuring IT alumni and training workshops for K-12 teachers. We hope to inspire other institutions to adopt our model, or adapt its components, to enhance their own outreach initiatives and BPC efforts.
