Aaltodoc Repository :: Browsing by Author "Hellas, Arto"

Browsing by Author "Hellas, Arto"

Now showing 1 - 20 of 101

Access management solution to corporate service
(2024-05-20) Hyytiäinen, Jami; Marjasalo, Joona; Perustieteiden korkeakoulu; Hellas, Arto
This master's thesis introduces identity and access management methods as part of IT security activities, examines how different identity and access management methods are used in a business service belonging to Pohjola Insurance Ltd, and explores how access management could be improved in the business service. Semi-structured interviews were organized to gain deeper insight on the issues and service requirements. To improve the access control in the service, the aim was to suggest applicable development approaches. The approaches were evaluated against the service requirements and additional evaluation criteria. A case study was performed to Pohjola Insurance Ltd, to study the reasons of why and manners of how the current access control solution is flawed in the service. The purpose was also to collect access control demands that would be required from an improved solution. Gathered results from the case study reveal that the current way of performing access control within the service is flawed mainly because access control was overlooked in the past and developed to keep up with the demanding business needs instead of staying ahead of them. Based on the evaluation, four different development approaches were proposed. Query-based approach could be one option, but it would require approval and collaboration from the database management teams. Access Control as a Service could be another option, although additional investigations would be needed to decide whether to build, buy or use an existing attribute-based access control platform as a base. Modularization of the access control should be studied more within the service context. It could be a valid option, if confirmed applicable. Data replication would not unfortunately be a suitable option for the business service.
Achievement Goal Orientation Profiles and Performance in a Programming MOOC
(2020-06-15) Polso, Kukka Maaria; Tuominen, Heta; Hellas, Arto; Ihantola, Petri; Department of Computer Science; Professorship Malmi L.; University of Helsinki; University of Turku
It has been suggested that performance goals focused on appearing talented (appearance goals) and those focused on outperforming others (normative goals) have different consequences, for example, regarding performance. Accordingly, applying this distinction into appearance and normative goals alongside mastery goals, this study explores what kinds of achievement goal orientation profiles are identified among over 2000 students participating in an introductory programming MOOC. Using Two-Step cluster analysis, five distinct motivational profiles are identified. Course performance and demographics of students with different goal orientation profiles are mostly similar. Students with Combined Mastery and Performance Goals perform slightly better than students with Low Goals. The observations are largely in line with previous studies conducted in different contexts. The differentiation of appearance and normative performance goals seemed to yield meaningful motivational profiles, but further studies are needed to establish their relevance and investigate whether this information can be used to improve teaching.
Algorithm Visualization and the Elusive Modality Effect
(2021-08-16) Zavgorodniaia, Albina; Tilanterä, Artturi; Korhonen, Ari; Seppälä, Otto; Hellas, Arto; Sorva, Juha; Department of Computer Science; Lecturer Sorva Juha group; Professorship Malmi L.; Lecturer Korhonen Ari group; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); Computer Science - Computing Systems (ComputingSystems); Lecturer Seppälä Otto group; Lecturer Hellas Arto group; Lecturer Korhonen Ari group
The modality effect in multimedia learning suggests that pictures are best accompanied by audio explanations rather than text, but this finding has not been replicated in computing education. We investigate which instructional modality works best as an accompaniment for algorithm visualizations. In a randomized controlled trial, learners were split into three conditions who viewed an instructional video on Dijkstra's algorithm, with diagrams accompanied by audio, text, or both. We find neither a modality effect in favor of the audio condition nor a verbal redundancy effect in favor of using only a single modality rather than both. Taken together with earlier research, our findings suggest that the modality effect is difficult to apply reliably and computing educators should not rush to integrate audio into visualizations in expectation of the effect. We discuss theoretical viewpoints that future research should attend to; these include alternative part-explanations of the modality effect and attention-based models of working memory, among others.
Automated Program Repair Using Generative Models for Code Infilling
(Springer, 2023) Koutcheme, Charles; Sarsa, Sami; Leinonen, Juho; Hellas, Arto; Denny, Paul; Department of Computer Science; Wang, Ning; Rebolledo-Mendez, Genaro; Matsuda, Noboru; Santos, Olga C.; Dimitrova, Vania; Lecturer Hellas Arto group; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); Lecturer Hellas Arto group; University of Auckland
In educational settings, automated program repair techniques serve as a feedback mechanism to guide students working on their programming assignments. Recent work has investigated using large language models (LLMs) for program repair. In this area, most of the attention has been focused on using proprietary systems accessible through APIs. However, the limited access and control over these systems remain a block to their adoption and usage in education. The present work studies the repairing capabilities of open large language models. In particular, we focus on a recent family of generative models, which, on top of standard left-to-right program synthesis, can also predict missing spans of code at any position in a program. We experiment with one of these models on four programming datasets and show that we can obtain good repair performance even without additional training.
Automatic Generation of Programming Exercises and Code Explanations Using Large Language Models
(2022-08-03) Sarsa, Sami; Denny, Paul; Hellas, Arto; Leinonen, Juho; Department of Computer Science; Lecturer Hellas Arto group; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); Lecturer Hellas Arto group; University of Auckland
This article explores the natural language generation capabilities of large language models with application to the production of two types of learning resources common in programming courses. Using OpenAI Codex as the large language model, we create programming exercises (including sample solutions and test cases) and code explanations, assessing these qualitatively and quantitatively. Our results suggest that the majority of the automatically generated content is both novel and sensible, and in some cases ready to use as is. When creating exercises we find that it is remarkably easy to influence both the programming concepts and the contextual themes they contain, simply by supplying keywords as input to the model. Our analysis suggests that there is significant value in massive generative machine learning models as a tool for instructors, although there remains a need for some oversight to ensure the quality of the generated content before it is delivered to students. We further discuss the implications of OpenAI Codex and similar tools for introductory programming education and highlight future research streams that have the potential to improve the quality of the educational experience for both teachers and students alike.
Biosignals reflect pair-dynamics in collaborative work: EDA and ECG study of pair-programming in a classroom environment
(2018-12-01) Ahonen, Lauri; Cowley, Benjamin Ultan; Hellas, Arto; Puolamäki, Kai; Department of Computer Science; Professorship Kaski Samuel; Professorship Puolamäki K.; Finnish Institute of Occupational Health; University of Helsinki
Collaboration is a complex phenomenon, where intersubjective dynamics can greatly affect the productive outcome. Evaluation of collaboration is thus of great interest, and can potentially help achieve better outcomes and performance. However, quantitative measurement of collaboration is difficult, because much of the interaction occurs in the intersubjective space between collaborators. Manual observation and/or self-reports are subjective, laborious, and have a poor temporal resolution. The problem is compounded in natural settings where task-activity and response-compliance cannot be controlled. Physiological signals provide an objective mean to quantify intersubjective rapport (as synchrony), but require novel methods to support broad deployment outside the lab. We studied 28 student dyads during a self-directed classroom pair-programming exercise. Sympathetic and parasympathetic nervous system activation was measured during task performance using electrodermal activity and electrocardiography. Results suggest that (a) we can isolate cognitive processes (mental workload) from confounding environmental effects, and (b) electrodermal signals show role-specific but correlated affective response profiles. We demonstrate the potential for social physiological compliance to quantify pair-work in natural settings, with no experimental manipulation of participants required. Our objective approach has a high temporal resolution, is scalable, non-intrusive, and robust.
Building a Framework for Testing Background Service of Desktop Applications
(2023-12-11) Jaakkola, Tuukka; Anderson, Lasse; Nurmonen, Veikko; Perustieteiden korkeakoulu; Hellas, Arto
This Master's thesis focuses on the test environment for the background service of the Effector desktop application, known as Effector Service. It is motivated by the recognition that the existing testing procedures for Effector Service might be insufficient. The objective of the thesis is to enhance the testing of Effector Service through improved automated testing methods. This thesis adopts the Design Science Framework. First identifying the need for improvements in the automated testing of Effector Service. There was a hypothesis that automated testing of Effector Service should include testing in an environment closely resembling the production environment. Requirements for enhancing the testing process are considered. A testing tool is developed to improve the testing of Effector Service. This tool is then evaluated to determine how effectively it fulfills the requirements and rectifies the identified problems. The tool developed in this research facilitates testing of Effector Service in an environment close to the production environment. This enhancement opens up the possibility of conducting types of tests, which were previously challenging. During the development of the tool, a critical bug was discovered in Effector Service that, if not identified, could have potentially made its way into production. The tool will help prevent such issues from occurring in the future. This finding supports the hypothesis made in problem identification. Following the principles of the Design Science Framework, this research resulted in the creation of a tool that improves the automated testing of Effector Service. By employing this tool, Effector Service is better prepared to meet the quality and reliability standards required in software development.
Canvas-based Browser Fingerprinting
(2022-12-16) Impiö, Jussi; Hellas, Arto; Perustieteiden korkeakoulu; Hyvönen, Eero
Comparing Code Explanations Created by Students and Large Language Models
(2023-06-29) Leinonen, Juho; Denny, Paul; Macneil, Stephen; Sarsa, Sami; Bernstein, Seth; Kim, Joanne; Tran, Andrew; Hellas, Arto; Department of Computer Science; Lecturer Hellas Arto group; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); University of Auckland; Temple University; Lecturer Hellas Arto group
Reasoning about code and explaining its purpose are fundamental skills for computer scientists. There has been extensive research in the field of computing education on the relationship between a student's ability to explain code and other skills such as writing and tracing code. In particular, the ability to describe at a high-level of abstraction how code will behave over all possible inputs correlates strongly with code writing skills. However, developing the expertise to comprehend and explain code accurately and succinctly is a challenge for many students. Existing pedagogical approaches that scaffold the ability to explain code, such as producing exemplar code explanations on demand, do not currently scale well to large classrooms. The recent emergence of powerful large language models (LLMs) may offer a solution. In this paper, we explore the potential of LLMs in generating explanations that can serve as examples to scaffold students' ability to understand and explain code. To evaluate LLM-created explanations, we compare them with explanations created by students in a large course (n ≈ 1000) with respect to accuracy, understandability and length. We find that LLM-created explanations, which can be produced automatically on demand, are rated as being significantly easier to understand and more accurate summaries of code than student-created explanations. We discuss the significance of this finding, and suggest how such models can be incorporated into introductory programming education.
Computing Education in the Era of Generative AI
(ACM, 2024-01-25) Denny, Paul; Prather, James; Becker, Brett A.; Finnie-Ansley, James; Hellas, Arto; Leinonen, Juho; Luxton-Reilly, Andrew; Reeves, Brent N.; Santos, Eddie Antonio; Sarsa, Sami; Department of Computer Science; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); Lecturer Hellas Arto group; University of Auckland; Abilene Christian University; University College Dublin; Lecturer Hellas Arto group
Challenges and opportunities faced by computing educators and students adapting to LLMs capable of generating accurate source code from natural-language problem descriptions.
Computing Education Research in Finland
(2023) Malmi, Lauri; Hellas, Arto; Ihantola, Petri; Isomöttönen, Ville; Jormanainen, Ilkka; Kilamo, Terhi; Knutas, Antti; Korhonen, Ari; Laakso, Mikko-Jussi; López-Pernas, Sonsoles; Poranen, Timo; Salakoski, Tapio; Suhonen, Jarkko; Department of Computer Science; Apiola, Mikko; López-Pernas, Sonsoles; Saqr, Mohammed; Computer Science Professors; Computer Science - Computing education research and educational technology (CER); Computer Science - Human-Computer Interaction and Design (HCID); Professorship Malmi L.; Computer Science Lecturers; Lecturer Hellas Arto group; Lecturer Korhonen Ari group; University of Helsinki; University of Jyväskylä; University of Eastern Finland; Tampere University; LUT University; University of Turku
Despite being a small country, Finland has been highly visible in international Computing Education Research (CER). This is demonstrated by the presence of several important research groups, dozens of graduated PhD students in CER during the last 20 years, and the success of the Koli Calling International Conference of CER, which has been running for 20 years now. In this chapter, we present the development of the CER field in Finland, the profiles of various research groups, and the roles of several national level networking activities which have supported the field. We discuss factors behind the strong presence and success of CER in Finnish universities.
Content Management Systems for Indoor Positioning
(2021-10-18) Pentakota, Likhitha; Hellas, Arto; Perustieteiden korkeakoulu; Ylä- Jääski, Antti
In recent days, we all know how the automation world affects our everyday lifestyle using navigation systems in ways that make their life more productive, safer, and more accessible. In recent research, Indoor Positioning System (IPS) has become one of the most trending topics. Effective IPS need a user-friendly UI paradigm to visualize and maintain the Indoor positioning data to enhance the user experience. A Literature review has been done on various Indoor Positioning Technologies and IPS applications that use the IPS technologies, including Wi-Fi, RFID, ultrasound, Bluetooth, Inertial Navigation System (INS), and Ultra-Wideband (UWB), and hybrid. In addition, a Literature review has also been done on existing CMSreal-time applications that are developed in research areas. A review has been done in the Analysis section on various Map libraries and their comparison with React Leaflet. Finally, a proposed web-based CMS has been implemented. It has three phases that include Authentication, BuildingManagement, and Floor Management. In Building management, the authenticated user can see the existing buildings on a map and a list of the buildings. Users can add or delete the buildings, and there will be an option to navigate to view the floor map or edit the floor map. In-Floor Management, the user can edit the building details, add floors, delete floors and draw blocks on each floor. In the proposed CMS solution, React has been used as front-end technology, React Leaflet is used for rendering maps, MongoDB is used for storing authenticated user’s information, and JSONServer is used for the backend to store the buildings and its floor data.
Creating a Machine Learning model to predict graduation
(2022-07-29) Järvinen, Joakim; Rytkönen, Anni; Perustieteiden korkeakoulu; Hellas, Arto
In this thesis, student data from Aalto University's Computer Science program was used to generate multiple binary classification machine learning models to predict whether students graduate or drop out. Model performance and related studies were analysed to create recommendations for the next steps in terms of data selection and model creation. The study also investigates how much data is required before predictions become accurate. From a total dataset of over 1000 students, around 500 students were selected for training and validation. The dataset consists of mainly academic features, such as course completions and grades, but a few personal characteristics such as nationality, gender and age were used as well. Models analysed were Logistic Regression, Naive Bayes, Support Vector Machines, k-Nearest Neighbors, Artificial Neural Networks and Random Forest. Training and validation were done by using kfold cross-validation and models' performance were analysed using a few common machine learning metrics such as accuracy, precision, recall, f1-measure, false positive rate and feature importance. Findings reveal that after 4 semesters of academic data, all models have adequate performance compared to a dummy classifier. Findings also reveal personal features to be relevant in all data sizes, and students' age to be one of the most important features when predicting graduation or drop-out. The thesis suggests that additional investigation is required to increase performance when the amount of academic data is low. The thesis recommends a more detailed investigation of how different socio-economic or personal features affect model performance, as well as improving the models using hyperparameter tuning methods.
Crowdsourcing Content Creation for SQL Practice
(2020-06-15) Leinonen, Juho; Pirttinen, Nea; Hellas, Arto; Department of Computer Science; Professorship Malmi L.
Crowdsourcing refers to the act of using the crowd to create content or to collect feedback on some particular tasks or ideas. Within computer science education, crowdsourcing has been used - for example - to create rehearsal questions and programming assignments. As a part of their computer science education, students often learn relational databases as well as working with the databases using SQL statements. In this article, we describe a system for practicing SQL statements. The system uses teacher-provided topics and assignments, augmented with crowdsourced assignments and reviews. We study how students use the system, what sort of feedback students provide to the teacher-generated and crowdsourced assignments, and how practice affects the feedback. Our results suggest that students rate assignments highly, and there are only minor differences between assignments generated by students and assignments generated by the instructor.
Crowdsourcing in Computing Education Research: Case Amazon MTurk
(2020-11-19) Hellas, Arto; Zavgorodniaia, Albina; Sorva, Juha; Department of Computer Science; Computer Science - Computing education research and educational technology (CER); Professorship Malmi L.; Computer Science Lecturers
Crowdsourcing platforms such as Amazon MTurk provide access to a human workforce that can be given tasks to complete online for a fee. In this article, we review studies in computing education research (CER) that rely on crowdsourcing; we also describe our own experiences of using Amazon MTurk for a CER study. We discuss challenges in recruiting workers with specific backgrounds—such as no programming experience—and considerations in filtering out unreliable research participants. Combining recommendations from the literature with the lessons that we learned whilst conducting our study, we synthesize advice for researchers in CER who are considering crowdsourcing. In our case study, we did not find widespread foul play by crowdsourced workers and, overall, our experiences and the literature suggest that crowdsourced CER is feasible. It is, however, uncertain to what extent crowdsourced data can produce answers that apply to specific educational contexts. More research is needed before definitive conclusions can be drawn about the validity and generalizability of crowdsourced CER.
Deadlines and MOOCs: How Do Students Behave in MOOCs with and without Deadlines
(IEEE, 2020-10-21) Ihantola, Petri; Fronza, Ilenia; Mikkonen, Tommi; Noponen, Miska; Hellas, Arto; Department of Computer Science; Lecturer Hellas Arto group; University of Helsinki; Free University of Bozen-Bolzano
Online education can be delivered in many ways. For example, some MOOCs let students to proceed with their own pace, while others rely on strict schedules. Although the variety of how MOOCs can be organized is generally well understood, less is known about how the different ways of organizing MOOCs affect retention. In this work, we compare self-paced and fixed-schedule MOOCs in terms of retention and work-load. Using data from over 8.000 students participating in two versions of a massive open online course in programming, we observe that drop-out rates at the beginning of the courses are greater than towards the end of the courses, with self-paced MOOC being more extreme in this respect. Mostly because of different starts, the fixed-schedule course has a better overall retention rate (45%) than its self-paced counterpart (13%). We hypothesize that students initial investment of time and effort contributes to their persistence in their course, meaning that they do not want to let their initial investment go to waste. At the same time, in both self-paced and fixed-schedule MOOCs, there are students who receive almost full points from one week but fail to continue to the next week. This suggests that the issue of dropouts in MOOCs may also be related to participants struggling to take up new tasks or schedule their work over a longer time period. Our results support scheduling student activities in open online courses and opens up new research directions in engaging students in self-paced courses.
Decades of Striving for Pedagogical and Technological Alignment
(2024-02-06) Haaranen, Lassi; Ahrenberg, Lukas; Hellas, Arto; Department of Computer Science; Lecturer Haaranen Lassi group; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); Professorship Vuorimaa P.; Lecturer Hellas Arto group
Computing educators have a tradition of building systems for supporting students and teachers alike. Systems and their evaluations are also an important topic in computing education research. Learning systems range from very specific such as algorithm visualizations to broad learning management systems, including supporting systems such as Q&A services. Over the years, calls for interoperability have also emerged, fueled by the desire for sharing best practices and content. In this discussion paper, we consider the challenges with evolving technology and aligning technology with pedagogy. We see that building and using systems is a learning effort that can help the community grow, while we also note that there are no perfect solutions and that there always will be trade-offs in developing and using technology in computing education.
Decoding Logic Errors: A Comparative Study on Bug Detection by Students and Large Language Models
(2024-01-29) MacNeil, Stephen; Denny, Paul; Tran, Andrew; Leinonen, Juho; Bernstein, Seth; Hellas, Arto; Sarsa, Sami; Kim, Joanne; Department of Computer Science; Computer Science Lecturers; Computer Science - Computing education research and educational technology (CER); Lecturer Hellas Arto group; Temple University; University of Auckland; Department of Computer Science
Identifying and resolving logic errors can be one of the most frustrating challenges for novices programmers. Unlike syntax errors, for which a compiler or interpreter can issue a message, logic errors can be subtle. In certain conditions, buggy code may even exhibit correct behavior - in other cases, the issue might be about how a problem statement has been interpreted. Such errors can be hard to spot when reading the code, and they can also at times be missed by automated tests. There is great educational potential in automatically detecting logic errors, especially when paired with suitable feedback for novices. Large language models (LLMs) have recently demonstrated surprising performance for a range of computing tasks, including generating and explaining code. These capabilities are closely linked to code syntax, which aligns with the next token prediction behavior of LLMs. On the other hand, logic errors relate to the runtime performance of code and thus may not be as well suited to analysis by LLMs. To explore this, we investigate the performance of two popular LLMs, GPT-3 and GPT-4, for detecting and providing a novice-friendly explanation of logic errors. We compare LLM performance with a large cohort of introductory computing students (n = 964) solving the same error detection task. Through a mixed-methods analysis of student and model responses, we observe significant improvement in logic error identification between the previous and current generation of LLMs, and find that both LLM generations significantly outperform students. We outline how such models could be integrated into computing education tools, and discuss their potential for supporting students when learning programming.
Design and Implementation of a Web Analytics process for a Gaming Site
(2022-06-13) Härkönen, Sami; Väisänen, Johanna; Perustieteiden korkeakoulu; Hellas, Arto
In this work, we concentrate to a Finnish gaming site which provides gaming services to millions of users who can access them by using either web application or mobile application. Thus, there is a lot of data coming from users’ actions which are gathered by using Google Analytics. We concentrate only to the organization’s mobile application which alone has millions of actions happening each day. However, the organization’s mobile development team lacks a clear way of monitoring the data and key statistics. Thus, we need to find relevant metrics which are important to the organization. After we have decided the key metrics, we need to decide how to report them such that they can provide insights to the organization’s product development. Web Anlytics can be used for understanding and optimizing the usage of web sites and mobile applications. It is not a technology but a process which have the following steps: define goals for the web site, define key performance indicators (KPIs) for measuring success of the objectives, collect data, analyze data by converting data into insights so that changes can be implemented. Otherwise, the KPIs are useless if no changes are made based on the insights. We are going to use constructive research process to design and implement a construction which is going to be a web anlytic process that solves the organization’s problem of deciding KPIs, finding KPIs and reporting KPIs. We managed to create a web analytic process which helped to determine important KPIs for the organization, process the data for the KPIs such that irrelevant data was excluded and only relevant data was included, and report the KPIs such that they can be used for gaining insights in the product development. Thus, this work illustrates that web analytic processes can be applied to mobile applications which resembles a typical web application.
Designing and Building a Platform for Teaching Introductory Programming supported by Large Language Models
(2024-01-22) Joy Kulangara, Kiran; Koutcheme, Charles; Perustieteiden korkeakoulu; Hellas, Arto
Large language models (LLMs) have the potential to improve programming education by providing feedback and guidance to students. Despite their potential benefits, the integration of LLMs into education presents unique challenges, including the risk of over-reliance on their feedback and the inconsistency of feedback quality. Addressing these concerns requires research to identify effective ways of integrating LLMs into programming education, which itself is challenging due to the rapid evolution of LLMs. To meet this challenge, this thesis introduces a flexible platform that can integrate multiple LLMs, providing an experimental space for research and innovative approaches to enhance programming education through LLMs. Guided by the Design Science Research Methodology framework, the thesis outlines the design, development, and evaluation of this educational platform. Conducted at Aalto University’s LeTech research group, the thesis presents an introductory programming learning platform specifically tailored to the group’s research objectives. The platform facilitates data collection, and enables the students to have a personalized learning experience with the help of LLM feedback. The work advances our understanding of LLMs in education and feedback mechanisms’ importance. The developed platform effectively demonstrates the feasibility of integrating LLMs into programming education. A small-scale study evaluating the platform’s overall usability received an average rating of 4.21 out of 5.00, while the LLM feedback received an average usefulness rating of 4.28 out of 5.00, highlighting its effectiveness and value in assisting students. Though the study sample size was small, the findings are encouraging. Future research could use the platform to explore multiple LLMs and conduct studies to improve the feedback mechanisms.

Browsing by Author "Hellas, Arto"

Results Per Page

Sort Options