Digital Library of Gesellschaft für Informatik e.V.
Not a member yet
40758 research outputs found
Sort by
"Don't roll (but build) your own crypto": Lessons from building a zk-STARK proving system
No full textKonzeption eines Integrations-Hubs in der Google Cloud zur Ablösung businesskritischer Systeme
No full textDie zunehmende Digitalisierung führt zu mehr Software-Nutzung in Unternehmen, wobei der Trend zu spezialisierten Services geht. Enterprise Application Integration (EAI) verbindet diese oft isolierten Systeme. Es wurde untersucht, ob businesskritische Systeme durch einen Integrations-Hub problemlos ablösbar sind. Hierfür wurden EAI-Strategien analysiert und ein Integrations-Hub in der Google Cloud mit Data Lake, Data Warehouse, Datenintegrations-Pipelines und Apigee X als API-Management-Plattform entwickelt. Das Mitarbeiter-Modul der abzulösenden Business Intelligence Lösung wurde durch eine interne API ersetzt. Die Ablösung und der testweise Austausch einer API zeigen, dass Software durch den Hub einfach und ohne Beeinflussung verbundener Systeme ersetzt werden kann. Die Architektur ermöglicht eine saubere Datenverarbeitung und ist erweiterbar
Comicification: Human-centered approach to data visualization and participant autonomy by redesigning surveys
No full textTraditional surveys, while essential for data collection in research, often struggle to maintain participant engagement, leading to low response rates and compromised data quality. This study explores an alternative approach: integrating comics and visual storytelling into survey design—a process we term “comicification.” Through a series of co-creation workshops, artists and researchers collaboratively transformed text-based surveys into visual narratives. Initial thematic analysis of the workshop process highlights how artists used images, character-driven storytelling, and interactive elements to make survey questions more intuitive, relatable, and engaging. While the project is ongoing, our process and initial findings are meant to provide provocation and inspiration for future "comicification" research
Bridging Emotional and Rational Dimensions: Developing A Process for Art-Based Education on Water
No full textRational information alone is insufficient to convey complex water issues; emotionally resonant experiences are essential to drive meaningful action. We propose a novel process, referred to the Perceptual Translation Process (PTP), that supports artists in transforming scientific data into immersive, participatory artworks. Grounded in the is–ought gap in environmental communication, the PTP consists of three integrated components: Data Embodiment, Narrative Immersion, and Action Catalysis. Through a case study in the Danube River region, we illustrate how the PTP can address pressing issues. This work offers a structured, yet flexible model for interdisciplinary collaboration among artists, scientists, and communities. Beyond environmental engagement, PTP has potential for broader applications in science communication, education, and public engagement across domains such as HCI, ubiquitous computing, and interactive media. Future work will focus on evaluating the real-world impact of PTP implementations and refining its applicability in diverse contexts
Study of robustness of Quantum Machine Learning and its countermeasures against adversarial attacks when facing noise
No full textQuantum computing and quantum machine learning are currently providing solid first results. It is only little time that needs to go by when these techniques and technologies will transfer from being emergent to facing application in our real-world industry use cases. This is also emphasized from the United Nations marking 2025 as the International Year of Quantum Science and Technology. Entering industry, we face not only the goal of accuracy and generalizability of the trained models, but also their robustness against cybersecurity attacks. Therefore, this study focuses on adversarial attacks in quantum machine learning (QML) from the early stages of this technology being used. We provide an overview of already and successfully conducted experiments for adversarial attacks in quantum machine learning (QML) using the Projected Gradient Descent (PGD) and Fast Gradient Sign Method (FGSM) attacks as well as using perturbations in training data to increase robustness as a countermeasure on simulator devices – therefore, facing no noise. We extend these experiments to be run on noisy simulators and present the robustness in training with and without perturbed training data
Ohne Umwege
No full textMit Sensoren und Drohnen Waldbrände vermeiden? Das klingt erst mal nach einer großen Errungenschaft. Aber wenn die Technik plötzlich auch zur Überwachung von Menschen genutzt werden soll, flammen schnell Diskussionen auf
Spatial Exploration of Melodic Topologies in Indian Classical Music using XR
No full textIndian classical music is based on a complex system of Raagas (melodic modes) and Talas (rhythmic cycles), where each Raaga evokes specific emotions. To explore the melodic structures in Indian classical music we propose a XR (eXtended Reality) based visualization. The visualization is driven by a guitar with MIDI pick-up. The spatial visualization of different Raagas facilitate the exploration of structural symmetries and tonal relationships inherent in Raagas for offering new opportunities for understanding, composition, and educational interaction
Driving Together: An Analysis of Passengers' Needs and Desire for Cooperative Control in Automated Vehicles
No full textDriving automation aims to enhance comfort, safety, and traffic flow by removing the human driver from the control loop. However, the human experience of commuting involves more than just reaching a destination or assuming the role of a driver. Factors like personal driving style and courtesy towards fellow road users are integral to the driving experience but often overlooked in the development of driving algorithms for automated vehicles. In this study, we explored the needs of passengers in highly automated vehicles. A qualitative use case analysis was conducted (N=16). In a second study, N=15 participants experienced the resulting use cases in an automated vehicle. In these scenarios, they were able to interact with the automation through a cooperation HMI. Results indicate that most participants expressed a desire for cooperative driving, albeit varying with the driving situation. Moreover, allowing cooperation improves passengers' overall experience by satisfying psychological needs for autonomy, security, competence, and relatedness
FAIR4EDTECH: Towards an Artifact Evaluation Track
No full textThis poster proposes a community-driven Artifact Evaluation Track for the DELFI conference to promote FAIR principles and recognize research artifacts such as software and data. Building on recent initiatives within the AK Open Science of the section Educational Technology of the German Informatics Society, we outline requirements and processes for submitting, evaluating, and awarding badges to artifacts. Emphasis is placed on domain-specific standards in Educational Technology, including suitable metadata ontologies and data exchange formats like xAPI. Review procedures should be time-bound, criteria-based, and ideally include interactive rebuttal phases. The goal is to foster transparency, reproducibility, and reusability in educational technology research by aligning with practices from established computer science venues. The poster invites discussion on how to implement and sustain such a track within DELFI
Evaluating Unstructured Log Parsing Techniques for Log Anomaly Detection
No full textModern computing environments generate logs for failure detection and causal analysis, with most anomaly detection methods beginning with log parsing to extract constant log event templates by removing variable components. However, changing log formats over an application's lifecycle makes traditional log parsing increasingly error prone. Recent methods bypass log parsing, using deep learning and NLP techniques to identify anomalies. This paper presents a comparative evaluation of two such state-of-the-art log anomaly detection approaches, supervised and self-supervised learning, represented by NeuralLog and LogFiT, respectively. Our experimental evaluation on popular benchmark datasets indicates that the supervised method achieves higher accuracy at the cost of lower operational efficiency. Finally, we discuss the implications of these results for practitioners and briefly explore a possible direction of future work that overcomes this trade-off