Fraunhofer Institute for Open Communication Systems

The transition toward decentralized energy systems has amplified interest in peer-to-peer electricity trading. However, research on prosumer behavior in such markets remains fragmented, hindered by a lack of benchmarkable experimental infrastructure. Addressing this gap, the LabChain system was developed—a modular, interactive prototype designed to study human behavior in synthetic P2P electricity markets under controlled laboratory conditions. This system integrates real-world technologies, such as blockchain-based transaction backends, flexibility market interfaces, and asset control tools, allowing fine-grained observation of strategic and perceptual dimensions of prosumer activity. The research followed an iterative design approach to develop the infrastructure for experimental energy economics research, and to assess its effectiveness in aligning participant experience with design intentions. Based on the meta-requirements generality, affordance-centric design, and technological grounding, 13 detailed peer-to-peer market, software, and system requirements that allow for system evaluation were developed. As a proof of concept, seven participants simulated prosumer behavior over a week through interaction with the system. Their interaction with the system was analyzed through simulation data and focus group interviews, using a modified thematic content analysis with a hybrid inductive–deductive coding approach. The main achievements are (i) the design and implementation of the LabChain system as a modular infrastructure for P2P electricity market experiments, (ii) the development of an associated experimental workflow and research design, and (iii) its demonstration through an illustrative, proof-of-concept evaluation based on thematic content analysis of a single focus group session focusing on interaction and perceptions. The behavioral results from an initial session are limited, exploratory, and demonstrative in nature and should be interpreted as illustrative only. They nevertheless revealed tension between system flexibility and cognitive usability: while the system supports diverse strategies and market roles, limitations in interface clarity and information feedback constrain strategic engagement.

Das vorliegende Kapitel beleuchtet die digitale Transformation des deutschen Staates in einem komplexen Mehrebenensystem, das von Bund, Ländern und Kommunen geprägt ist. Trotz umfangreicher Gesetzesinitiativen wie dem Onlinezugangsgesetz (OZG) bleibt die Umsetzung vieler digitaler Verwaltungsleistungen hinter den Erwartungen zurück. Koordinationsprobleme, föderale Zuständigkeiten und historisch gewachsene Strukturen erschweren eine einheitliche Vorgehensweise. Um den Rückstand aufzuholen, wurden zentrale Institutionen wie der IT-Planungsrat und die FITKO geschaffen, die bund- und länderübergreifende Standards und Prozesse vorantreiben sollen. Gleichzeitig macht sich eine starke Abhängigkeit von externen IT-Dienstleistern bemerkbar, was den Aufbau interner Kompetenzen verzögert. Erfolgsgeschichten wie die Digitalisierung des Bundesamts für Migration und Flüchtlinge oder kommunale One-Stop-Shops verdeutlichen jedoch, dass nutzerzentrierte Ansätze und agile Methoden effektiv sein können. Neue Formate wie Digitalisierungslabore und Netzwerke wie NEXT bündeln Expertise aus verschiedenen Verwaltungsebenen, um innovative Lösungen zu entwickeln. Der Beitrag zeigt, dass technologische Fortschritte allein nicht genügen: Nur ein kultureller Wandel in der Verwaltungsarbeit, unterstützt durch Schulungen und umfassende Koordination, kann langfristig zu einer bürgerfreundlichen und effizienten digitalen Verwaltung führen.

AI implementation is part of the digital transformation efforts of National Libraries. To understand what the challenges are and derive recommendations for other libraries, we researched the AI implementation. Based on 90 interviews with library and AI experts, we identified ten European National Libraries and included the Library of Congress and the British National Library in our case analysis. Here, we derive the challenges that National Libraries face when implementing AI technologies. Based on these insights, we provide recommendations for libraries that plan to also implement AI solutions in their organisations. This chapter focuses on sustainable governance structures that ensure long-term strategic alignment of funding and implementation goals, the promotion of technical independence of international providers through internal competence and upskilling efforts, and the involvement of information specialists and users to ensure adoption and acceptance of AI tools. Furthermore, we highlight the importance of collaborative and interdisciplinary approaches as well as transparency, responsible data management, and ethical AI.

AI implementation in National Libraries is a multi-actor endeavour that includes many internal as well as external contributors. The literature on the subject is, however, sparse and still developing. In this chapter, I provide the theoretical framework and the research design for the EU LibrarIN project that is built on the co-production literature in the public management field. The resulting theoretical framework of digital co-production has guided the case study selection and inquiry and set the foundation for the empirical data analysis of the 12 case studies on AI implementation in National Libraries.

The Royal National Library of Belgium (KBR) collects all Belgian publications and safeguards the country’s cultural and historical heritage. Given the increasing volume of records and the existence of records that are only catalogued on paper cards, there emerged a need for an AI-based solution. In response, KBR initiated a project called Retrocatalography, collaborating with the external service provider Initum. This partnership resulted in the development of an AI-based cataloguing solution which was created in just 10 days. Built on the Microsoft Power Platform, this low-code application is capable of gathering information from a book’s cover, archiving it, and verifying the details. In addition, KBR developed the Allez app in-house to scan library records. Volunteers use the app by assisting in the verification process to ensure accuracy of the records data. The AI-based solution went live in October 2022, with strong support from top management of the library. Along the way, KBR encountered several challenges, including staff shortages, time constraints for integrating the system into daily operations, and various technical difficulties. Overall, the AI-based solution transforms the workflow of librarians at KBR. It has made cataloguing books more efficient, ultimately enhancing the quality of the library services.

The Library of Congress’ (LoC) project “Exploring Computational Description” (ECD) is investigating the use of machine learning (ML) to create metadata for e-books that have not yet been catalogued. In-house the LC Labs carried out this initiative with the U.S. Programs, Law, and Literature Division and an external vendor. An initial budget of $250,000 from the National Digital Trust Fund was allocated for this experimental AI endeavour, prompted by a massive backlog of e-books. During the first project phase, five ML models were evaluated, and in the second project phase, human-in-the-loop prototypes that offer machine-generated terms to librarians were introduced. The integration of AI at the LoC has the potential to enhance cataloguing efficiency by automating repetitive tasks, thereby allowing librarians to focus more on intellectual tasks. At the same time, the project faced several challenges, including ensuring the reliability of AI-generated records, copyright concerns, and managing potentially harmful language in older texts used for training the models. Improving the accuracy of these models remains essential and depends on access to extensive digital data. However, human expertise remains crucial for ensuring high quality, and librarians need to develop a foundational understanding of ML to leverage these technologies effectively. The aim of the project is to develop innovative approaches that contribute to improving library practices.

As part of its commitment to innovation and to improve the accessibility of the library's collection of resources from the long nineteenth century, the British Library ran the "Living with Machines" project between 2018 and 2023. The Library's project uses Artificial Intelligence (AI) technologies to improve access to historical collections. The library faced the challenge of making the extensive collections easier to interact and more accessible to users. The project was initiated to explore how AI and machine learning (ML) would impact historical research, with a particular focus on the impact of mechanisation in the nineteenth century. The project team was gathered from 2018 to July 2019. The project involved over 5500 volunteers and engaged the public through crowdsourcing tasks. The collaboration spanned various disciplines, including data scientists, historians and research software developers. The British Library also reached out to the public to ask for their help in commenting on the artefacts. Collaboration challenges arose when staff did not spend enough time understanding each other's disciplines. According to the interviewees, librarians will need a solid grounding in "information organisation" when it comes to working with AI solutions and outputs.

The National Library of Finland developed and implemented an automated subject indexing and classification tool, Annif, which is now also used by other National Libraries. The tool is integrated into Finto AI, an in-house metadata management system launched in 2020. These two AI-driven solutions were created by the growing volume of digitally collected data, the extensive digitisation of paper archives, inadequate AI support for various Finnish languages, and a lack of transparency in the services offered by external providers on the market. The successful implementation of Annif and Finto AI can be attributed to the strong collaboration between the project team and their external stakeholders, such as other National Libraries or universities, as well as the awareness within the Finnish National Library. Use cases, for example, from the University of Jyväskylä, highlight the value of these AI-driven solutions. However, challenges remain, such as identifying hidden errors in the automatically generated records. Fortunately, the National Library's extensive in-house expertise allows for the immediate detection of mistakes in both the code and the generated records.

The German National Library (DNB), as mandated by the German National Library Act of 2006, is responsible for collecting all German cultural artefacts, including digital and non-physical resources. To meet this requirement, DNB engaged in a Machine Learning (ML) project to improve metadata generation. After a while the project transitioned to a completely automated approach called subject cataloguing machine using Annif, an AI tool from the National Library of Finland, incorporating ML or subject indexing and metadata creation. This AI development relies heavily on the expertise of librarians, as AI tools are at times not fully accurate. While AI enhances efficiency, librarians and cataloguing specialists remain essential to ensure quality. The transition faced challenges, including limited resources, ethical issues, a diversified workforce’s attitude towards AI, and the complexities of copyright laws. Despite these challenges, the integration of EMa has improved the generation and accessibility of metadata within DNB services. While AI can enrich librarians’ skills, it cannot replace them. Mistakes encountered along the way should be viewed as learning opportunities, and future advancements should involve collaborative, diverse teams to continue refining the integration of AI in metadata management.

Cerebral blood flow regulation is critical for brain function, and its disruption is implicated in various neurological disorders. Many existing models do not fully capture the complex, multiscale interactions among neuronal activity, astrocytic signaling, and vascular dynamics, especially in key brainstem regions. In this work, we present a 3D-1D-0D multiscale computational framework for modeling the neuro-glial-vascular unit (NGVU) in the dorsal vagal complex (DVC). Our approach integrates a quadripartite synapse model, which captures the dynamic interactions among excitatory and inhibitory neurons, astrocytes, and vascular smooth muscle cells, with a hierarchical description of vascular dynamics that couples a three-dimensional microcirculatory network with a one-dimensional macrocirculatory representation and a zero-dimensional synaptic component. By linking neuronal spiking, astrocytic calcium and gliotransmitter signaling, and vascular tone regulation, our model reproduces key features of neurovascular regulation and elucidates the feedback loops that help maintain cerebral blood flow. Simulation results demonstrate that neurotransmitter release triggers astrocytic responses that modulate vessel radius, thereby influencing local oxygen and nutrient delivery. This integrated framework provides a robust and modular platform for future investigations into the pathophysiology of cerebral blood flow regulation and its role in autonomic control, including the regulation of gastric function.

As the availability of historical biodiversity data continues to grow, ensuring its usability through adherence to FAIR principles (Findable, Accessible, Interoperable, and Reusable) has become increasingly essential. This study focuses on solving key challenges in interpreting biodiversity data from historical texts, particularly in identifying and aligning common species names with their modern scientific counterparts. We address five main challenges: spelling variations, the invention of new terms, semantic shifts between broad and narrow naming conventions, and the renaming or reclassification of historical terms. To tackle these issues, we tested a range of large language models (LLMs) (GPT‑4, LLaMA3-405B, Mistral-8B, and Qwen3-30B-A3B) for their ability to resolve these challenges and support terminology alignment. The initial entity detection was performed using GPT-4o, which achieved a 92% success rate in detecting historical common names and correctly identified 98% of scientific terms on a test dataset. Comparative evaluation of the ability to match historical common names with modern equivalents revealed that GPT-4o consistently delivered the most accurate and nuanced outputs across four of the five challenges, demonstrating strong contextual understanding. The results highlight the potential of advanced LLMs to not only identify entities but also to interpret historical naming conventions, thereby enhancing the reusability and interoperability of biodiversity data in line with FAIR principles.

Lieferketten und -netzwerke sowie deren Entitäten und Beziehungen lassen sich als Wissensgraphen semantisch erfassen und darstellen, um sie sowohl für Menschen als auch Maschinen nutzbar zu machen. Die häufige Unvollständigkeit von Informationen über Lieferketten erfordert es, Hypothesen zu formulieren und in die Analyse einzubeziehen. Auf dieser Grundlage wurde das ViSCA-System entwickelt, das eine kartenbasierte, grafische Erfassung von Lieferketten und zugehörigen Hypothesen benutzerfreundlich unterstützt. Seine graphische Analysekomponente ermöglicht die Berechnung von Transportrouten, die Ermittlung von Transportdauern und kritischen Transportwegen, die Identifikation von potenziellen Engpassknoten für redundante Auslegungen sowie die Abschätzung der Produktverfügbarkeit unter Berücksichtigung multipler potenzieller Disruptionen.