Browse all accepted contributions. Each entry shows the paper title, authors, and full abstract. * = presenting author.
Introducing Programming Concepts through Montessori Materials
Steffi Rudel (University of the Bundeswehr Munich)*; Marlene Engels (University of the Bundeswehr Munich)
Programming is a skill that is already taught in many schools today. However, for historical reasons, it has not yet been included in the curriculum of Montessori schools. To close this gap, we developed a learning materials for Montessori schools and present it in this article.
We show the design of a learning material for introducing programming concepts in Montessori primary school. The material is a concrete physical learning material and is supported by a modular lesson structure. Also we discuss its suitability for Montessori education based on expert feedback.
The article shows interesting examples of how programming can be taught at the Montessori primary level.
Summative Assessment of Program Comprehension in CS1 with Questions about Learners' Code
Afonso Caniço (Iscte - Instituto Universitário de Lisboa)*; André Santos (Iscte - Instituto Universitário de Lisboa)
Questions about Learners' Code (QLCs) assess to what extent learners understand their own code through personalized questions that target previously written code. We conducted an experiment involving 18 students of an Introductory Programming (CS1) course taught in Java, comparing the scores of their midterm test based on code-writing questions with the scores of a multiple-choice test based on QLCs. The questions targeted student code submitted throughout the semester in an automated assessment system. Regression analysis shows a strong association between the scores of both tests, suggesting comparable discriminatory power. Participants generally reacted positively and recognised the benefits of this kind of assessment. Our results suggest that QLC tests may serve as a valid complement for CS1 summative assessment targeting program comprehension.
Exploiting Generative AI to Scale Up Intelligent Tutoring Systems
Miguel Soares Ferreira (DCC - FCUP)*; José Paulo Leal (DCC - FCUP)
The research described in this paper explores the integration of Generative Artificial Intelligence (GenAI) into Intelligent Tutoring Systems (ITS) with the aim of improving programming education. Specifically, it extends Agni, a web-based programming learning platform, by automating the construction of key ITS components that are traditionally built manually. The methodology leverages Large Language Models (LLMs) to extract programming concepts from educational materials, map their relationships via concept graphs and diagnose specific student knowledge gaps. Additionally, GenAI provides real-time, contextual feedback during programming exercises, mimicking the personalized support typically offered by a human tutor. The work evaluates whether GenAI can effectively replace or augment the manual creation of domain models while preserving the pedagogical benefits of ITS. The obtained results suggest the potential of combining AI with ITS to improve scalability and reduce manual workload.
Algorithm X: Competitive programming as a teaching methodology for Algorithms and Data Structures in high school
Odair Moreira de Souza (Federal Institute of Education, Science and Technology of Paraná (IFPR) and Federal University of Paraná (UFPR))*; Clodis Boscarioli (Western Paraná State University of (Unioeste)); Letícia Mara Peres (Federal University of Paraná (UFPR))
Teaching algorithms and data structures in Technical High School involves challenges related to abstraction, problem solving, and the persistence of non-student-centered teaching models. Competitive Programming and Competition-Based Learning emerge as promising pedagogical alternatives. This study aims to analyze the educational impacts of the teaching project “Algorithm X” on students’ learning of algorithms and data structures, as well as on their motivation, autonomy, academic engagement, and perceptions regarding Competitive Programming in Technical High School. The study is based on the experience of the teaching project, developed with students from the Technical Course in Informatics Integrated with High School at IFPR — Campus Cascavel. The project was organized through weekly meetings structured around dialogic expository classes, problem solving, coding dojo, debates, simulated contests, and digital support resources. The data were obtained through an evaluative questionnaire, and discursive responses were analyzed. The sample comprised 37 students who participated in the project in 2025. The results indicated a favorable evaluation of the project and recognition of its contribution to strengthening algorithmic thinking, developing problem-solving skills, motivating the study of programming, improving performance in programming-related subjects, and fostering learning autonomy. The project was understood as a space for learning and development extended beyond the immediate context of competitions. Competitive programming is considered a viable pedagogical strategy for teaching algorithms in technical high schools when supported by structured didactic mediation, progressive practice, and digital resources, consolidating technical skills and competencies and contributes to students' academic and professional development. Student participation in the project was not limited to the context of competitions but also involved a complementary formative process.
Gamification in Programming Education: A Zelda-Inspired RPG Approach for Reinforcing Foundational Programming Skills
Max Fritsche (DHSN / Security Robotics GmbH)*
The acquisition of programming skills is a notoriously difficult process characterized by high cognitive load, complex abstraction requirements, and significant dropout rates in higher education. Traditional pedagogical approaches often struggle to bridge the gap between seemingly abstract language syntax and the strategic problem-solving required in software engineering. This paper presents a comprehensive gamified solution: a 2D top-down Role-Playing Game (RPG) inspired by classical adventure games such as The Legend of Zelda, specifically designed to reinforce existing knowledge. We detail the integration of a secure, Docker-based automated code judge within a narrative-driven environment. Key features include a virtual economy, a social “company” system for collaboration, and a curriculum focused on the C programming language to deepen low-level understanding. The paper evaluates the technical robustness of the prototype and proposes an empirical framework to validate the hypothesis that this immersive approach significantly enhances learning productivity, persistence, and student motivation.
Architecture and design study with analytical validation of a mixed reality software system for pilot training with real-time visual scene substitution
Serhii Prykhodchenko (Dnipro University of Technology)*; Oksana Prykhodchenko (Dnipro University of Technology); Andrii Martynenko (Dnipro University of Technology)
This paper presents the design and architectural formalization of a mixed reality (MR) system for pilot training based on selective, region-level visual scene substitution. Unlike conventional VR/AR/MR approaches that operate on entire scenes or additive overlays, the proposed paradigm enables context-aware replacement of semantically meaningful regions within the pilot’s egocentric view—such as cockpit displays and external windows—while preserving direct interaction with physical controls and the pilot’s hands.
The core contribution is a modular real-time architecture that integrates semantic perception, structured scene representation, and GPU-accelerated mask-based compositing into a unified pipeline. The system is formulated as a dataflow-driven process with explicitly defined stages and dependencies, enabling analytical reasoning about latency, computational complexity, and performance bottlenecks under strict real-time constraints (below 20–30 ms). A formal mask-based compositing model supports fine-grained, region-specific manipulation of the visual environment.
Adopting a design-science perspective, the work introduces a structured evaluation framework that combines analytical validation, prototype-level technical assessment, and a pathway toward empirical studies. This addresses a key limitation in XR-based aviation training research, where rigorous analytical grounding prior to large-scale human evaluation is often insufficient.
By combining high sensorimotor fidelity with controllable visual variability, the proposed approach defines a new class of hybrid training systems that bridge physical cockpits and fully simulated environments, providing a foundation for future adaptive MR systems in safety-critical domains such as aviation.
Learning, Not Just Coding
Necmi Sapoglu (University of British Columbia Okanagan)*
AI coding assistants such as GitHub Copilot and ChatGPT are highly effective at generating and correcting code, but their optimization for productivity raises challenges in educational contexts, where immediate solutions can bypass the reasoning processes necessary for learning. This paper presents VibeLearner, a Visual Studio Code extension that constrains AI assistance through a Role–Task–Requirements–Instructions (RTRI) prompting framework. The system supports three interaction modes, Socratic, Hinted, and Show-and-Tell, that regulate how assistance is delivered to promote guided reasoning, reflection, and incremental problem solving. A synthetic, scenario-based comparison with an unconstrained AI assistant illustrates how interaction-level constraints influence the structure and pedagogical character of responses. This work contributes a design framework for embedding scaffolded AI behavior into development environments and highlights the role of interaction design in aligning AI assistance with learning-oriented objectives.
From Student Code to Adaptive Learning: A Personalized System for Programming Education
Michaela Ďurkovičová (TUKE); Marek Horváth (TUKE)*; Lenka Bubeňková (TUKE); Emília Pietriková (TUKE)
Personalized learning is one approach to addressing differences in students’ prior knowledge, pace, and programming practice. This paper presents the design and implementation of a web application that supports personalized learning with an emphasis on the individual needs of students. The system design is based on an analysis of existing solutions and technological approaches, the results of which were incorporated into the system architecture. The proposed application enables instructors to create and manage tests through an administrative interface while providing insights into student activity and performance. Students can complete predefined tests as well as initiate the generation of personalized tests based on selected parameters. The application includes a module for generating personalized programming challenges, which analyzes the student’s source code obtained from a university version control system and evaluates selected code quality metrics. Based on this analysis, it generates tasks aimed at addressing identified weaknesses. After completing the assignment, the application provides automated evaluation along with textual feedback. The outcome of this work is a functional system prototype, which was tested with students at an anonymized technical university. The evaluation focuses on usability, task completion, and perceived acceptance rather than on direct measurement of learning gains. The system is designed with extensibility and scalability in mind, enabling further evaluation in broader university environments.
Bridging Game Development and Virtual Reality Education through Blender-Based Tutorials
Lenka Bubenkova (Technical University of Košice)*; Marek Horvath (Technical University of Košice)
The integration of 3D modeling tools into STEM education can improve creativity, technical competence, and learner engagement. This study evaluates the use of Blender-based tutorials in two university courses: Game Development (GD) and Systems of Virtual Reality (SVR). Over four semesters, data were collected from 420 students using anonymous questionnaires and course performance indicators. Results indicate increased student confidence in using Blender, positive perceptions of relevance and satisfaction, and improved understanding of core modeling concepts across both courses. Students in the SVR course demonstrated higher engagement, while GD students more frequently applied self-created assets in practical projects. These findings suggest that Blender can serve as an effective educational bridge between software engineering, game development, and immersive technologies.
Inverted Grading in Project-Based Learning: A Learning Path Approach for Sustainable Performance Evaluation
Filipe Portela (University of Minho)*
Higher education assessment is traditionally based on cumulative grading models, where students progressively build their final classification over time. However, such approaches often promote risk-averse behaviour and limit continuous improvement.
This paper proposes an inverted grading strategy integrated with learning paths within the TechTeach paradigm. In this model, students are evaluated according to the maximum expected grade of their selected path and assessed on their ability to maintain or recover their performance throughout the course. The strategy combines group and individual evaluation, enabling effective differentiation of student contributions in project-based learning environments.
The approach was validated through a case study conducted in a Web Programming course with 153 students. The results show a dynamic grade evolution, with an average decrease from 16.31 at the first control point to 15.34 at the second, followed by a recovery to 16.05 in the final evaluation. The overall variation (-0.26, or -1.6%) indicates that final outcomes remain close to initial performance despite intermediate fluctuations. Additionally, 62% of students experienced a grade decrease, 28% improved their performance, and 10% maintained stable results, confirming the model’s ability to capture performance dynamics and support recovery.
Student feedback further reinforces the approach's applicability, with more than 80% of responses indicating acceptance of the model.
Overall, the results suggest that inverted grading provides a robust alternative to traditional cumulative models, promoting performance sustainability, accountability, and continuous engagement in higher education.
Codula, a Social Network Geared Towards Programming Education
João Rodrigues (UMinho); Alvaro Costa Neto (IFSP)*; Alice Balbé (UMinho); Pedro Henriques (UMinho)
Digital social networks have become an ubiquitous presence to almost everyone, to the point that they are frequently taken for granted whenever someone gets asked for their contact. Beyond that, it is without doubt that their capacity for stimulating participation in groups and exchange of ideas, while retaining attention for long periods of time is—for good, or for bad—highly efficient. If taken for a good cause, their features can become fierce allies to educators in their teaching efforts. Specially in the case of Computer Programming, a sharp, and difficult matter, social networks can be of great value, not only for their inherent technological link to Computer Programming, but also for their natural connection to students that dare to pursue the arduous path to become programmers. Codula is a social network that was designed and implemented with the purpose of paving and shortening this path, both for students, and educators alike. This paper presents both Codula's design principles that guided its construction, and main features, that were carefully crafted to fulfil its purpose. While foundational concepts solidified design choices, such as the absence of different roles for users, and the stimulus of the sense of belonging via personal customisation, features were implemented to directly support Computer Programming Education, such as the specification of different post types, and functionalities that facilitate composing, sharing, and commenting source code. By converting opposing forces of a commonly found distraction, Codula aims to leverage the best aspects of social networks to support a better cause, in the hopes of stimulating a successful collective teaching-learning process.
Catching What the Borrow Checker Can’t: Towards Serious Game-Based Security Training for Rust Developers
Frederico d'Abreu (ISCTE-IUL)*; Tiago Gasiba (Siemens AG); Sathwik Amburi (Siemens AG); Maria Albuquerque (ISCTE-IUL)
Cybersecurity incidents cause significant financial damage to organizations, making secure software development a critical industry priority. Although Rust offers strong memory safety guarantees, developers can still introduce security vulnerabilities through improper coding practices. In this paper, we present the preliminary design of a serious game for secure Rust coding training in an industrial setting. We investigate real-world Rust vulnerabilities, map them to CWEs, and propose challenge scenarios alongside a structured evaluation protocol for validation by industry security experts. Our design incorporates defense-oriented coding tasks, a three-level scaffolded hint system, and AI-based backend evaluation, grounded in established pedagogical principles such as situated learning and cognitive load management. This work provides a foundation for future implementation and empirical evaluation of learning outcomes with professional software developers.
Digital Educational Resources in Programming Education: A Systematic Review of Resource Types and Teaching Challenges
Ana Paula Silva Paulina Bezerra (PPGCC UERN UFERSA); Ceres Germanna Morais (Uern)*; Gabryella Rocha Rodrigues (IFPA)
Teaching programming requires the use of different digital educational resources to support the understanding of abstract concepts and the development of problem-solving skills. However, these resources are often scattered across multiple platforms, which makes it difficult for teachers to select, evaluate, and reuse them. This article presents a Systematic Literature Review on the use of digital educational resources in programming education, focusing on the types of resources employed, teaching practices, and reported challenges. The review was conducted based on a structured protocol, including the definition of search strings, the selection of databases, and the application of inclusion and exclusion criteria. After applying these criteria, 27 primary studies were selected for the analysis phase. The results reveal a predominance of practice-oriented resources, such as programming exercises and interactive environments, together with the use of multimedia materials and reusable content. Despite this diversity, significant challenges persist, particularly regarding resource selection, pedagogical integration, material quality, and teachers’ workload. These findings highlight the need for more structured approaches to support teachers in the effective use and organization of digital educational resources in programming education.
The SOB Monitor: Supporting Competency-Based Assessment with Gamification
David Dorvekinger (Middlesex University)*; Michael Heeney (Middlesex University); Kelly Androutsopoulos (Middlesex University)
The SOB Monitor is a web-based platform that supports competency-based assessment through Student Observable Behaviours (SOBs). SOBs are defined as measurable units aligned with learning outcomes and are assessed at Threshold, Typical, and Excellent levels. The assessment strategy involves continuous assessment with multiple resubmission opportunities, allowing for real-time monitoring of student progress. It is adopted on several undergraduate Computer Science modules, including first-year Programming, Systems and Architecture, Foundations of Computing, and Design and Development of Applications.
In this paper, we describe assessment with SOBs, the new architecture of the SOB Monitor platform, its implementation and how it supports competency-based assessment in Computer Science undergraduate modules. Gamification elements including leaderboard and rankings have been introduced to enhance student motivation, engagement and self-regulated learning within pass/fail modules.
5W2H as a Flexible Analytical Framework for Investigating Programming Teaching and Learning Across Different Methodological Designs
Gabryella Rodrigues (Universidade do Minho)*; Ceres Morais (UERN - Universidade Estadual do Rio Grande do Norte); António Osório (Universidade do Minho)
Teaching and learning programming in Higher Education is a complex process involving cognitive, pedagogical, contextual, and metacognitive dimensions. Traditional analyses focused only on final learning outcomes, such as code correctness, performance, or pass/fail rates, may not fully capture how students construct knowledge, face difficulties, mobilize strategies, and interact with different learning contexts. This article discusses the use of the 5W2H framework as a flexible analytical tool in two doctoral studies on programming teaching and learning. The first study used 5W2H to organize and triangulate data from a mixed-methods single case study on the teaching and learning of programming in a Higher Education course. The second study articulated 5W2H with the spiral model and Bloom's revised taxonomy to analyze a qualitative multiple case study focused on students' algorithm construction process. Through a comparative analysis, the article shows that 5W2H can operate at different analytical scales: as a broad structure for mapping subjects, contexts, methodologies, difficulties, and learning factors, and as a process-oriented framework for interpreting students' actions during algorithmic problem solving. The findings suggest that 5W2H supports the organization, categorization, triangulation, and interpretation of complex educational data. Therefore, the framework offers a useful methodological lens for research in programming education, helping researchers understand the relationships between subjects, contexts, strategies, difficulties, and cognitive processes involved in teaching and learning programming.
Ogma: An Intelligent Platform for Anticipating Academic Failure and Recommending Recovery Measures
Ricardo Queirós (ESMAD, Polytechnic of Porto)*
The growth of class sizes, the increasing complexity of course content, and the multiplicity of digital platforms make it progressively harder for instructors to identify, in a timely manner, students who are at risk of academic failure. This short paper presents the concept of Ogma, a learning analytics application with artificial intelligence components designed to anticipate signs of academic decline and suggest personalized recovery measures. The proposal combines heterogeneous data - Moodle interactions, forum messages, grade records, assignment submission dates, student-produced work, and logs from other educational applications - to build a dynamic risk profile. Based on this profile, the system provides instructors with a dashboard containing graded alerts, explanations of risk factors, and pedagogical recommendations generated by a pipeline that combines machine learning, retrieval-augmented generation, and large language models. The paper presents the motivation for the problem, synthesizes related work, and describes the design and implementation of a reference architecture for Ogma.
Bridging Generative AI and Gamification in Moodle: Design and Evaluation of GamiBot
José Carlos Paiva (P. Porto); Ricardo Queirós (ESMAD, Polytechnic of Porto)*; Raquel Almeida (CIR, ESS, Polytechnic of Porto, Portugal); Teresa Terroso (ESMAD, Polytechnic of Porto); Mário Pinto (ESMAD, Polytechnic of Porto)
While Learning Management Systems (LMS) like Moodle are ubiquitous in higher education, they often lack dynamic, personalized academic support. We present GamiBot, an intelligent pedagogical agent natively integrated into Moodle to bridge this gap. GamiBot combines Large Language Model (LLM) assistance, Retrieval-Augmented Generation (RAG) grounded in course materials, adaptive quizzes, and a structured gamification backend. To evaluate its effectiveness, we conducted a two-week pilot study across two higher education institutions. Results from telemetry (773 interactions) and student surveys demonstrated high system usability (74.6% positive sentiment) and strong chatbot performance (69.4%). Students heavily favored active self-regulation, dedicating 53% of interactions to generating adaptive quizzes rather than passive content summaries. Conversely, gamification elements received mixed feedback (31.7%), underscoring the need for technical refinement and high customizability. Ultimately, GamiBot suggests that integrating generative AI directly into existing LMS workflows can provide reliable, context-aware formative support that aligns with self-regulated learning strategies.
No papers match your search.