This paper introduces a novel benchmark dataset designed to evaluate the capabilities of Vision Language Models (VLMs) on tasks that combine visual reasoning with subject-specific background knowledge in the German language. In contrast to widely used English-language benchmarks that often rely on artificially difficult or decontextualized problems, this dataset draws from real middle school curricula across nine domains including mathematics, history, biology, and religion. The benchmark includes over 2,000 open-ended questions grounded in 486 images, ensuring that models must integrate visual interpretation with factual reasoning rather than rely on superficial textual cues. We evaluate thirteen state-of-the-art open-weight VLMs across multiple dimensions, including domain-specific accuracy and performance on adversarial crafted questions. Our findings reveal that even the strongest models achieve less than 45% overall accuracy, with particularly poor performance in music, mathematics, and adversarial settings. Furthermore, the results indicate significant discrepancies between success on popular benchmarks and real-world multimodal understanding. We conclude that middle school-level tasks offer a meaningful and underutilized avenue for stress-testing VLMs, especially in non-English contexts. The dataset and evaluation protocol serve as a rigorous testbed to better understand and improve the visual and linguistic reasoning capabilities of future AI systems.

Virtual reality has proven to be a valuable addition in the tool belt of teachers. Immersive learning environments are applied in various settings, including, but not limited to the medical and nursing domain. In this study we present “We care in VR”, a simulation for practicing nursing tasks for care at home, a part of nursing that is currently underrepresented in available VR applications. We investigate how realistic interactions are perceived by end users compared to consistent usage of buttons on the controllers and how they affect the ease of use of the simulation. We conduct an empirical study with 50 participants from three vocational schools of nursing and a university of applied sciences. Results suggest that our simulation already works quite well and is accepted by the target group, but still needs improvement regarding ease of use, especially for users without any previous experience with VR applications.

This study examines how Large Language Models (LLMs) can reduce biases in text-to-image generation systems by modifying user prompts. We define bias as a model's unfair deviation from population statistics given neutral prompts. Our experiments with Stable Diffusion XL, 3.5 and Flux demonstrate that LLM-modified prompts significantly increase image diversity and reduce bias without the need to change the image generators themselves. While occasionally producing results that diverge from original user intent for elaborate prompts, this approach generally provides more varied interpretations of underspecified requests rather than superficial variations. The method works particularly well for less advanced image generators, though limitations persist for certain contexts like disability representation. All prompts and generated images are available at https://iisys-hof.github.io/llm-prompt-img-gen/

This paper considers three fundamental approaches to software development, namely manual coding, modeldriven software engineering, and code generation by large language models. All of these approaches have their individual pros and cons, motivating the desire for an integrated approach. We present MoProCo, a technical solution to integrate the three approaches into a single tool chain, allowing the developer to split a software engineering task into modeling, prompting or coding sub-tasks. From a single input file consisting of static model structure, natural language prompts and/or source code fragments, Java source code is generated using a two-stage approach. A case study demonstrates that the MoProCo approach combines the desirable properties of the three development approaches by offering the appropriate level of abstraction, determinism, and dynamism for each specific software engineering sub-task.

Similar to LLMs, the development of vision language models is mainly driven by English datasets and models trained in English and Chinese language, whereas support for other languages, even those considered high-resource languages such as German, remains significantly weaker. In this work we present an analysis of open-weight VLMs on factual knowledge in the German and English language. We disentangle the image-related aspects from the textual ones by analyzing accuracy with jury-as-a-judge in both prompt languages and images from German and international contexts. We found that for celebrities and sights, VLMs struggle because they are lacking visual cognition of German image contents. For animals and plants, the tested models can often correctly identify the image contents according to the scientific name or English common name but fail in German language. Cars and supermarket products were identified equally well in English and German images across both prompt languages.

Dieser Band rekapituliert empirische Projekte der Forschungsgruppe "Empirical Research and User Experience (ERUX)“ am Institut für Informationssysteme (iisys) der Hochschule Hof aus den Jahren 2023 bis 2024.

Im Fokus steht die Usability von Features im Fahrzeuginnenraum, nämlich Fahrerassistenzsysteme, Optionen für die Klimaanlage und Situational Awareness Interfaces.
Während im ersten Band dieser Reihe neben einer großzahligen Erhebung vor allem die apparative Beobachtung mittels Eyetracking Anwendung fand, verwenden die im aktuellen Band referierten Forschungsprojekte ein eher qualitatives
empirisches Methodenspektrum, von Expertenratings über die Analyse von Use Cases bis zu Konzepttests mit zahlenmäßig begrenzten Usergruppen. Daneben ist eine Conjoint-Studie enthalten.
Im Automobilsektor hat die „Markt“- oder „Kundenorientierung“ in den letzten Jahrzehnten kaum Fortschritte gemacht, die Zufriedenheit der Kunden mit dem Automobil ist eher rückläufig. Im Zeitablauf haben zuerst die Kosten-
einsparungswelle und nachfolgend die Erfordernisse und die Präferenzen der Softwareentwicklung die HMI im Fahrzeuginnenraum viel stärker determiniert als die Usability für die Autofahrer, die das fertige Produkt letztlich bezahlen müssen.
Und weiterhin haben bei einigen OEM mächtige Designabteilungen das letzte Wort bei allen Oberflächen, mit denen der Autofahrer in Kontakt kommt. Als Fazit einer Vielzahl von Detailerkenntnissen aus den hier berichteten Studien lässt
sich feststellen, dass für die Berücksichtigung von Useranforderungen in der Entwicklung weiterhin viele Ansatzpunkte bestehen. Zugleich zeigt sich, dass die Auskunftsfähigkeit von Usern in Bezug auf zukünftige Fahrsituationen begrenzt ist.

Dieser Band redet somit keineswegs einer einseitigen Dominanz „des Marketings“ oder „der Marketingforschung“ im Entwicklungsprozess das Wort. Richtig bleibt vielmehr, was schon seit Jahrzehnten bekannt ist: Es kommt darauf an, ingenieurbasierte Innovations-
leistungen, kreative Designentwürfe, kundenseitige Nutzenerwartungen und marktgegebene Kostenvorgaben in ein ausgewogenes Verhältnis zu bringen. Das Know-how über den dafür erforderlichen, laufenden Abstimmungsprozess entscheidet über den Markterfolg

This volume recaps selected empirical projects from the research group Empirical Research and User Experience (ERUX) at the Institute for Information Systems (iisys) at Hof University from 2023 to 2024. The focus is on the usability of features inside the vehicle, specifically driver assistance systems, climate control options, and situational awareness interfaces, which are designed to provide drivers with orientation and safety when transitioning from manual to automated driving modes. While the first volume of this series included a large-scale survey and primarily relied on apparatus observation using eye tracking, the research projects referenced in the current volume employ a more qualitative empirical method spectrum, ranging from expert ratings to the analysis of use cases and concept testing with numerically limited user groups. As in the first volume, a conjoint study is also included. In the automotive sector, the focus on "market" or "customer orientation" has made little noticeable progress in recent decades, and customer satisfaction with automobiles is declining. Over time, the cost-saving wave and, later, the demands and preferences of software development have determined HMI in the vehicle interior much more strongly than usability from the perspective of the drivers, who ultimately pay for the final product. Furthermore, at some OEMs, powerful design departments still have the final say on all surfaces that the driver interacts with. As a conclusion from a variety of detailed findings, it can be stated that there are still many areas for the consideration of user requirements in development. At the same time, the studies reported here show that users' ability to provide meaningful feedback on future driving situations and associated requirements is limited. This volume does not advocate for the one-sided dominance of "marketing" or "marketing research" in the development process. Rather, what has been known for decades remains true: It is essential to strike a balance between engineering-based innovations, creative design concepts, customer expectations of value, and market-driven cost requirements. The expertise required for the ongoing coordination process is what ultimately determines market success.

This study explores the integration of Industry 5.0 principles into existing Industry 4.0 configurations to address excessive heat dissipation, energy efficiency and promote carbon neutral manufacturing. Industry 5.0 represents a paradigm shift, emphasizing human-centered approaches, energy efficiency and environmental sustainability alongside traditional productivity goals. The case study analyzes energy consumption in an Industry 4.0 setup, quantifies heat waste and applies innovative solutions to optimize energy use. Considering digital manufacturing through the Industrial Internet of Things (IIoT), big data analytics, smart automation technologies in cyber-physical production systems and sustainable practices, the study aims to mitigate and minimize energy waste and carbon emissions. Through quantitative analysis, it identified energy inputs, mapped energy transformations, and assessed energy waste hotspots. Despite all the benefits of Industry 4.0, it showed that responsible practices and new initiatives are required to reuse and capitalize on that wasted energy or heat, underscoring the importance of integrating Industry 5.0 principles for sustainable manufacturing.

Die großen digitalen Plattformen bestimmen nicht nur die Regeln auf ihren Märkten, sondern wirken zunehmend auch auf die gesellschaftliche Sphäre ein, mit oft als äußerst schädlich empfundenen Resultaten, wie etwa der Schaffung von Filterblasen und „Rabbit Holes“ oder der unkontrollierten Verbreitung von Fake News. Immer stärker werden daher die Rufe nach einer Beschränkung der Plattformmacht, nach Plattformregulierung oder sogar Zerschlagung der dahinterstehenden Konzerne laut. Dabei stellt sich die Frage, ob entsprechende Maßnahmen nicht an anderer Stelle ansetzen sollten – etwa beim „Treibstoff“ der Plattformökonomie, den Daten.

Geht man davon aus, dass die Macht der großen Tech-Plattformen ihren Ursprung vor allem in der Sammlung von Daten und deren algorithmischer Auswertung hat, so sollte sich womöglich auch die Plattformregulierung entsprechend hieran orientieren. Während der erste Teil dieses Artikels sich mit den Grundlagen der Plattformmacht im „Überwachungskapitalismus“ beschäftigte, widmet sich dieser zweite Teil nun möglichen Alternativen zu den gegenwärtigen Regulierungsbemühungen. Diese setzen nicht bei den gegenwärtigen Marktverhältnissen, sondern bei den individuellen Datenverfügungsrechten an.