This dissertation investigates creativity and collaboration across different levels of virtuality, including remote, hybrid, and in-person environments.
It applies a user-centered Design Science Research approach, incorporating qualitative techniques such as diary studies, expert interviews, and prototype assessments to analyze performance satisfaction, preferences, and team dynamics.
The academic contribution features a conceptual framework that identifies and prioritizes creativity drivers, while practical design principles are refined and tested through remote and hybrid design thinking sessions. The results indicate that supporting creative processes involves adapting environments to match task needs, based on divergent and convergent thinking. Social and environmental affordances must balance social presence, anonymity, and accountability to maintain creative momentum within virtual teams. Additionally, a functional environment combined with effective technology management is key to fostering a flow state for creativity. As a practical outcome, recommendations for improving idea sharing and development in remote and hybrid teams assist organizations in navigating digital transformation and sustainably enhancing innovation processes.
Photovoltaics are a key technology in the transition toward renewable energy. Among emerging materials, poly(3-hexylthiophene) (P3HT) offers an attractive alternative to silicon owing to its solution processability, mechanical flexibility, and tunable optoelectronic properties. This thesis investigates how electric fields, mechanical deformation, and processing conditions influence the structural and photophysical properties of P3HT and the flexible substrate poly(ethylene terephthalate) (PET) using complementary time and frequency domain spectroscopic techniques. Femtosecond transient absorption pump-probe spectroscopy demonstrates that below-band-gap excitation directly generates polaron pairs (PPs) in P3HT. The transient response comprises fast PP and slower delocalized polaron pair (DPP) contributions. External electric fields modify the competition between DPP recombination and bimolecular annihilation, while uniaxial stretching up to 7% redistributes PP and DPP populations through polymer chain alignment. Above this strain, the transient response becomes increasingly complex owing to efficient strain transfer and microcrack formation. Raman and ultraviolet–visible spectroscopy correlate mechanical deformation with structural and optical changes. P3HT and PEDOT:PSS/P3HT films retain an essentially constant optical band gap up to approximately 7% strain, followed by a slight widening at 10%, indicating the onset of electronically significant deformation. Analysis of the Raman C=C and C–C stretching modes shows that variations in band position and full width at half maximum distinguish temperature-induced morphological changes from mechanically induced chain alignment. PET exhibits increased optical absorption and irreversible Raman spectral changes above approximately 5% strain, indicating molecular reorganization.
This thesis investigates how uniaxial tensile strain affects the optical and molecular properties of polymer thin films used in flexible organic electronics. It focuses on poly(ethylene terephthalate) (PET) as a flexible substrate and poly(3-hexylthiophene-2,5-diyl) (P3HT) as a semiconducting polymer. The materials are examined as single layers and in PET/P3HT and PET/PEDOT:PSS/P3HT stacks at room temperature.
Strain is applied reproducibly using a custom-built stretching setup. Ultraviolet–visible spectroscopy is used to track changes in optical response and band gap, while Raman spectroscopy reveals chain reorientation, conformational changes, and defect formation during and after deformation.
For PET, tensile strain causes progressive optical degradation, including increased absorption across the ultraviolet and visible regions. Above ≈5% strain, the changes become irreversible and are accompanied by evidence of permanent structural modification. This identifies two regimes: elastic behavior at low strain and irreversible deformation at higher strain.
For P3HT, the optical band gap remains stable up to 7% strain, regardless of annealing temperature or stack configuration. At 10% strain, a reproducible increase of about 4–5 meV appears, indicating onset of measurable electronic perturbation.
Raman results show that P3HT strain response depends strongly on thermal history. Unannealed films mainly deform through reversible chain alignment. Films annealed at 50 °C show a mixed response with greater microstructural heterogeneity, while films annealed at 75 °C exhibit more constrained deformation and stronger defect accumulation. A PEDOT:PSS interlayer improves strain transfer, reduces strain localization, and enhances mechanical tolerance, especially in highly ordered films.
Overall, the work defines experimentally grounded strain limits for flexible devices and clarifies how processing history and stack architecture govern mechanical and optoelectronic behavior.
This dissertation examines how digital technologies can support the integration of discipline-specific language in accounting and taxation. Across three essays, it analyzes the acquisition, application, and analysis of specialized terminology. Essay I investigates a wiki-based collaborative glossary in introductory accounting education and shows that it improves students' academic performance and engagement with accounting terminology, while also revealing challenges related to coordination and content reliability. Essay II uses dictionary-based text analysis to create a cross-country dataset on corporate environmental tax legislation and finds that such taxes are shaped more by institutional and political factors than by environmental risks. Essay III develops and applies a domain-specific large language model to qualitative corporate tax disclosures, demonstrating that specialized LLMs outperform general-purpose models and traditional text analysis methods. Overall, the dissertation shows that digital technologies can enhance the acquisition, application, and analysis of specialized language in accounting and taxation. It contributes to accounting education, tax policy research, and methodological debates on the use of natural language processing in discipline-specific contexts.
Children with Autism Spectrum Disorder (ASD) often experience difficulties in sensory processing, motor coordination, behavioral regulation, and adaptive functioning, which can limit participation in daily activities such as eating and social interaction. Selective eating is highly prevalent and may lead to nutritional and developmental challenges.
The Schmetterling Program was developed as an integrative, sequential intervention targeting selective eating, motor skills, and stress regulation in children with ASD. Inspired by the concept of the “butterfly effect,” the program assumes that small, individualized therapeutic changes can produce meaningful developmental improvements. It combines established behavioral strategies such as shaping, imitation chaining, and guided modeling within a sensory-motor framework.
This dissertation evaluates the effectiveness of the Schmetterling Program through three studies involving children aged 2–6 years with ASD. The first study examined the Nutritional Behavior Intervention (NBI) across three families and found improvements in dietary flexibility and adaptive behavior. The second study compared 24 children with a control group and showed reductions in selective eating, improvements in autism symptom severity, and enhanced autonomic regulation measured by heart rate variability (HRV). The third study assessed the Motor Treatment (MT) component using a multiple-baseline design and demonstrated improvements in motor coordination, sensorimotor integration, and social communication.
Overall, the findings suggest that the Schmetterling Program may improve eating behavior, motor functioning, and psychophysiological regulation in children with ASD. The results support the effectiveness of integrative, individualized, and ecologically grounded interventions. Further research is required to confirm long-term outcomes and underlying neurophysiological mechanisms.
(In)security in Africa influences African societies, geopolitics, and geoeconomics. Yet African perspectives on (in)security are rarely sought, and how African states shape their security policies remains unclear: conventional internal/external, domestic/foreign, and secure/insecure distinctions are outdated, while Western/Northern-centric social sciences and African Studies/IR literatures remain trapped in essentialisms, stereotypes, and mono-causal explanations. This thesis argues for new analytical devices to study African security policies in less essentialist and more context-sensitive ways by understanding the practice turn and appropriating Pierre Bourdieu's reflexive, practice-theoretical, and field-analytical sociology through an in-depth sociospatial reading. This helps interpret his concepts as global, interrelated thinking tools for African social formations, against a superficial Bourdieusian International Studies literature. Taking Nigeria as exploratory case study, the thesis traces how field and habitus intersect, develops a Field of Nigerian Security Policy (FNSP) as object and tool, and treats habitus and capital primarily as outcomes of the analysis. It asks how Nigeria shapes its security policy towards Boko Haram as a contemporary space of (in)security, drawing on almost four months of self-reflexive, ethnographic fieldwork. It combines sociological discourse analysis of textual sources with interviews with leading Nigerian experts and practitioners. It reveals how a FNSP is constituted by diverse agents, relations of (informal) cooperation, competition, domination, and transversality, a dual habitus of problems/scarcities and aspirations/necessities, and multiple (including negative) kinds of capital—and how agents' classificatory practices ultimately produce Nigeria's security policy towards Boko Haram. The thesis thereby advances African International Relations theoretically and offers deep practical insights into Nigeria's political space.
The study of pits, skylights, and sinkholes on terrestrial bodies is vital for mapping subsurface voids, geological evolution, and potentially habitable zones. Feature identification remains constrained by sensor resolution and the complexities of radar analysis. This research explores deep learning computer vision to automate landform mapping and evaluates detecting subsurface voids using orbital radar.
To address mapping challenges, this work introduces DeepLandforms, a toolkit developed to automate detection using You Only Look Once (YOLO), Detectron 2, and the Segment Anything Model (SAM). Built on Docker, it includes modules for data preparation, training, and inference, generating outputs compatible with GIS software. Validation against a dataset from the Mars Global Cave Candidate (MGC³) catalog demonstrates capacity for consistent, large-scale surveys.
This research also introduces EchoTerraeTrace, an all-in-one toolkit providing workflows for SHARAD (Mars Reconnaissance Orbiter) and MARSIS (Mars Express) data. During validation North-West of Ascraeus Mons, three unmapped volcanic vents were identified, and the regional paleotopography was refined.
Finally, this research assesses planetary software against FAIR principles. A web-based service built on a dockerized JupyterHub was developed as an all-in-one environment for standardized data processing. This service provides scalability without requiring local high-end resources, supporting reproducible research by moving the code to the data rather than the data to the code.
In intelligent and autonomous systems, users must often share private data with external platforms to enable advanced functionalities. A key challenge in this process arises when the data to be protected is not a single, precise value, but is instead inherently uncertain or represents a range of possible values. This thesis addresses the privacy of such uncertain data, which is formally represented and processed using set-based methods. We review existing privacy-preserving techniques and introduce new mechanisms to safeguard this data during processing on untrusted platforms, exploring both cryptographic and non-cryptographic approaches.
Specifically, the thesis presents novel privacy-preserving mechanisms for set-based data processing, categorized by the data type within the sets.
For sets of real-valued data, we introduce a differential privacy mechanism for set-based estimation in linear and non-linear dynamical systems. This approach protects the sensitive information contained within sets, such as zonotopes that model system uncertainties, while minimizing the loss of utility for the estimation process.
For sets of discrete, binary data, we propose a mechanism that uses Fast Fully Homomorphic Encryption to ensure privacy. This method allows for the secure processing of sets represented as logical zonotopes in untrusted environments, maintaining computational practicality.
The evaluations demonstrate the effectiveness and practicality of the proposed mechanisms across various applications in autonomous and intelligent systems.
Understanding how collective patterns emerge on graphs is a fundamental challenge across disciplines, from biological and ecological networks to computational and physical systems. This thesis explores the interplay between network topology and emergent dynamics using minimal models and spectral graph techniques.
A first investigation focuses on network inference, showing that Turing patterns encode structural information about the underlying graph, which we use to infer missing links. The second study investigates multistability in reaction-diffusion networks, showing how local spectral gaps influence the attractor landscape of Turing patterns using a heuristic binary classification algorithm. Finally, the third study applies the sandpile model to soil erosion processes, bridging concepts from self-organised criticality and connectivity-based geomorphology to investigate the role of minimal models in empirical research.
This thesis combines theoretical analysis, computational modelling and empirical validation to highlight how structure shapes dynamics across different contexts and illustrate the potential of minimal models as predictive, explanatory and exploratory tools for complex systems.
Rare earth elements (REE, or REY including yttrium) are widely used in modern technologies and are increasingly released into aquatic environments. Their environmental behaviour and bioaccumulation in aquatic ecosystems remain poorly understood. This thesis investigates the bioavailability, bioaccumulation, and trophic transfer of both geogenic and anthropogenic REY using aquatic organisms and environmental samples from European freshwater and marine systems.
Shells of three invasive freshwater bivalves (Corbicula fluminea, Dreissena polymorpha, and Dreissena bugensis) collected from seven major European rivers show strong REY bioaccumulation, with concentrations up to five orders of magnitude higher than in ambient water. Anthropogenic lanthanum contamination from the Rhine River was recorded in mussel shells, whereas no enrichment of anthropogenic gadolinium from MRI contrast agents was observed, suggesting its stability in freshwater systems.
Further analyses of freshwater (Anodonta anatina) and marine (Mytilus edulis) mussels reveal higher REY concentrations in internal organs than in muscle tissues and shells, while biological processes exert only minor influence on REY fractionation. A trophic-level study along the Rhine River shows a general biodilution trend from primary producers to fish, while shale-normalised REY patterns remain consistent across trophic levels. These results indicate that mussels can serve as effective biomonitors for environmental REY contamination.
The Influence of Digital Transformation on well-being – analysis of life stages and business sectors
(2026)
The accelerating pace of digital transformation (DT) is profoundly reshaping the world of work, placing new demands on employees and affecting their well-being. As employee well-being is closely linked to engagement and performance, this PhD project investigates how organizations can engage employees during DT, with particular consideration of their well-being. The Self-Determination Theory (SDT) serves as the kernel theory in this research for understanding well-being, expanded to include physical health. Furthermore, both different working conditions and various life stages of employees are incorporated in order to capture the dynamic nature of well-being. However, promoting well-being requires a comprehensive understanding of its multifaceted effects, both positive and negative, on employees, a challenge further intensified by the ongoing DT. While many companies recognize the benefits of DT, they often struggle with its implementation and the associated impacts on the workforce. Maturity models are a common tool to provide guidance during DT by serving as frameworks for assessing and developing organizational capabilities. In practice, maturity models are often too strategic, inflexible, and insufficiently user-centered. Furthermore, social aspects such as employee well-being have so far been largely neglected. To close this gap, an adaptable human-centered maturity model focusing on well-being was designed and empirically validated within the framework of this cumulative dissertation consisting of six papers, following the Design Science Research (DSR) approach. The model uniquely integrates basic psychological needs, physical health, and life stage perspectives, dimensions largely absent in existing DT maturity models. Overall, this PhD project advances the human-centered discourse on well-being by providing a practice-oriented maturity model that supports organizations in identifying the effects of DT on well-being and deriving appropriate courses of action.
This thesis investigates the integration of sustainability assessment considering Industry 4.0 technologies and the use of Markov Decision Process capabilities. The manufacturing industry is facing increasing pressure to improve sustainability assessment performance, and Industry 4.0 technologies like Digital Twins, Internet of Things, Big Data Analytics, Cloud Computing, Machine Learning, and Artificial Intelligence have the potential to support these efforts. However, effectively integrating sustainability assessment goals and Industry 4.0 technologies within manufacturing systems can be challenging. The research addresses this challenge by developing a framework for optimizing the flow of operations in a manufacturing system while incorporating sustainability assessment and Industry 4.0 technologies effectively. The framework utilizes the Markov Decision Process to model the decision-making process of the manufacturing system and its decision-makers. From the other side, it includes sustainability assessment goals as constraints or objectives in the Markov Decision Process model. The use of Industry 4.0 technologies is integrated into the framework to gather data and optimize the decision-making process based on that data. The thesis begins by reviewing the literature on sustainability assessment, Industry 4.0 technologies, and their impacts with regard to manufacturing systems. The proposed framework is then presented, and its capabilities are demonstrated through case studies of single and multiple agents on a shop floor. The trend in pioneer manufacturing firms is to implement new technological applications on their shop floor to agile their Manufacturing Execution System. The findings from the case study indicate that the proposed framework can effectively support decision-making at the top-tier level of the enterprise by integrating sustainability assessment and the Industry 4.0 paradigm.
This research paper investigates the advanced Artificial Intelligence (AI) architecture underpinning Germany's multimodal transportation ecosystem, drawing from the author's year-long immersive professional experience. The study shifts the focus from traditional physical logistics to the efficiency of data processing throughput, characterizing the transport network as a dynamic information organism.
The author analyzes the integration of Edge Computing and IoT sensors through the MQTT protocol, facilitating resilient data transmission in unstable environments. Furthermore, the paper details the implementation of Real-time Stream Processing using Apache Kafka and Redis, alongside the application of LSTM (Long Short-Term Memory) neural networks for high-precision delay forecasting. A significant technical analysis is provided on Neural Text-to-Speech (NTTS) synthesis for passenger notifications, emphasizing its role in enhancing user experience (UX) through natural language generation.
Beyond technical frameworks, the author addresses critical infrastructure security via TLS 1.3 and PKI, ensuring compliance with GDPR standards. The paper concludes that the success of modern transport lies in its "algorithmic soul"—a data-driven approach that prioritizes reliability and transparency. Ultimately, the author advocates for the strategic transfer of these AI-integrated architectures to Kazakhstan’s "Smart City" initiatives, suggesting that such digital transformation will serve as a catalyst for the broader technological evolution of the national economy.
Modern online courses often replicate traditional instruction as static artifacts, failing to reveal the cognitive causes of learner errors. This paper proposes a self-improving educational ecosystem integrating interactive modules, diagnostic assessments, and AI-driven analytics in a closed feedback loop.
The model is implemented on a real platform using WordPress as a flexible application framework. Each module combines theory, interactive practice (H5P), and diagnostic assessment. Natural-language queries to an AI assistant serve as diagnostic signals, revealing hidden cognitive barriers. A three-level management model separates operational support (AI tutor), pedagogical quality assurance, and strategic product development. Continuous improvement follows a four-stage cycle: signal collection, pattern analysis, targeted instructional adjustments, and impact verification.
This approach demonstrates that intelligent, evidence-based learning management can transform courses into self-correcting systems, where each cohort improves the experience for the next.
Digital transformation (DT) fundamentally reshapes organizational structures and work processes. Despite its strategic importance, up to 70% of DT initiatives fail, primarily due to insufficient consideration of human factors. This cumulative dissertation addresses this gap by developing and validating a human-centered Change Management Maturity Model that systematically integrates employee needs into digital transformation processes, with particular emphasis on the AI-driven third phase of DT.
Existing DT maturity models predominantly focus on technological, strategic, and organizational aspects while neglecting human-centered dimensions such as employee motivation, psychological well-being, and change readiness. Likewise, established change management frameworks tend to operate either at the organizational level (e.g., McKinsey 7S) or the individual level (e.g., ADKAR), without systematically integrating both perspectives. To address this limitation, this dissertation proposes a comprehensive maturity model comprising nine dimensions across three categories: Motivation & Leadership Behavior, Dealing with Change, and Well-being & Health.
The research follows an echeloned Design Science Research (eDSR) approach and is structured as a cumulative dissertation consisting of six research papers. The model is grounded in multiple kernel theories, including Self-Determination Theory, Herzberg’s Two-Factor Theory, Maslow’s Hierarchy of Needs, the Dynamic Capabilities Framework, and established change management models. Empirical validation was conducted in the skilled trades sector and across industries in the retail sector, demonstrating the model’s applicability across organizational contexts and its practical relevance for managing AI-driven transformation initiatives.
Digital tools are increasingly used in inclusive early childhood education and care (ECEC) to support children’s social skills — especially communication, emotion recognition, self-regulation, prosocial behavior, and peer interaction. Across Europe, policy momentum for inclusive digital education is accelerating, while research is expanding from general “screen time” debates toward evidence-based, developmentally appropriate, educator-mediated designs. This paper synthesizes current trends, highlights the European and German context, and proposes a feasible mixed-method study plan to evaluate digital social-skills interventions in inclusive ECEC settings. The review maps tool categories (tablet apps, serious games, social robots, multimodal platforms, and digital assessment/screening), describe equity and accessibility design principles, and identify evidence gaps (long-term outcomes, implementation fidelity, child-led vs. adult-guided interaction, and inclusion of multilingual/migrant families). We propose a pragmatic evaluation framework aligned with European priorities for accessibility, quality, and teacher capacity-building.
This paper examines current mathematical modeling methods used for the development and optimization of composite polymer stabilizers. A review of key challenges related to the stability of polymer materials is provided, along with a discussion of modern computational approaches, including molecular dynamics, finite element analysis, thermodynamic modeling, and machine learning. The necessity of an interdisciplinary approach integrating chemistry, materials science, and computational technologies is justified. Perspectives on the further development of modeling methods to enhance the efficiency and stability of polymer stabilizers are presented.
The research essay examines the opportunities and barriers to implementing artificial intelligence (AI) in the education system of the Republic of Kazakhstan based on a comparative analysis of the experience of Germany. The aim of the work is to identify key factors influencing the successful integration of AI tools into educational practice, including regulatory, methodological, technological and personnel aspects. The paper analyzes international and national strategies for regulating AI, features of the legal framework of Kazakhstan and the European Union, as well as models of teacher training. Special attention is paid to the issues of academic integrity, ethics, personal data protection and overcoming the digital divide. Based on the German experience, we formulate systematic recommendations for Kazakhstan aimed at developing institutional teacher training, reforming the assessment system and forming national AI sovereignty in education. The forecast of AI development in the educational system of Kazakhstan until 2029 is made, emphasizing the need to move from fragmented technology implementation to a sustainable, ethically verified and inclusive model of digital education.
This essay examines the role of artificial intelligence tutors in Kazakhstan's education system, addressing whether AI-based tools replace teachers or serve as supportive instruments that enhance teaching practice. Against the backdrop of ongoing digital transformation in education, the central thesis argues that AI tutors do not replace teachers but function as complementary tools that assist with specific tasks such as assignment evaluation, progress monitoring, and personalized feedback. While AI systems offer significant advantages — including personalized learning pathways, expanded access to educational resources in underserved rural areas, and reduced administrative burden for teachers — they cannot substitute essential pedagogical elements such as human interaction, emotional intelligence, cultural awareness, and moral guidance.
The analysis draws on theoretical frameworks from UNESCO and OECD, examining both the potential and limitations of AI integration within Kazakhstan's educational context, where regional disparities in infrastructure and digital literacy create uneven implementation. Through comparative insights from Germany's education system, the essay demonstrates how AI tutors can be effectively integrated as supportive tools while preserving the teacher's central role in pedagogical decision-making and student development. The findings emphasize that successful AI integration requires maintaining human oversight, addressing the digital divide, and ensuring that teachers retain professional autonomy in shaping educational outcomes.
This paper examines the integration of artificial intelligence (AI) into the education system, focusing on its key opportunities and challenges in the context of the Fourth Industrial Revolution. Artificial intelligence has become an important tool for modernizing education, enhancing learning quality, supporting personalized instruction, and improving educational management processes.
The study is based on a comparative analysis of the experiences of Germany and Kazakhstan in implementing AI in education. The German model emphasizes strategic planning, teacher training, and ethical and legal regulation, while Kazakhstan’s approach focuses on accessibility and rapid implementation through widely used EdTech platforms.
The findings indicate that the effective use of artificial intelligence depends not only on technological infrastructure but also on teacher readiness, data security, and ethical responsibility. The paper highlights the potential of AI to transform education and identifies the key conditions for its sustainable and balanced integration.
Underwater robots need reliable perception for navigation, mapping, diver interaction, and manipulation, yet vision is degraded by wavelength-dependent attenuation, scattering, and variable water optics. These effects reduce contrast, distort color, and destabilize visual cues, so perception must be tailored to underwater image formation and field reliability constraints.
This thesis develops multimodal, 3D-aware perception for adverse marine and deep-sea conditions, based on experiments and integration within the EU projects MORPH, CADDY, and DexROV. By combining complementary sensors (2D imagery, stereo 3D structure, inertial and acoustic cues) with learning pipelines, the approaches compensate for individual sensor weaknesses.
First, it enriches 2D perception with 3D context and underwater-specific enhancement. Contributions include terrain-complexity estimation from texture metrics and stereo geometry to adapt AUV speed during surveys, plus color restoration/image enhancement to improve detection and pose estimation. For human-robot interaction, it introduces diver detection and pose estimation that merge stereo point-cloud descriptors with recurrent neural networks to handle low-contrast imagery.
Second, it presents end-to-end systems, including the CADDY underwater stereo-vision dataset for gesture-based communication and a gesture-recognition pipeline that blends classical learning, deep detectors, and a grammar-guided human-in-the-loop design for safer diver and AUV communication.
Finally, for deep-sea intervention, it proposes a simulation-in-the-loop validation to reduce sim-to-real gaps and an adaptive localization framework fusing dense 3D reconstruction, planar geometry, image-quality cues, and visual odometry to maintain accurate navigation in low visibility. The methods are validated on real data and integrated into autonomous demonstrators for safety-critical missions during field trials.
This technical report provides a comprehensive analysis of the process of digital transformation in the education system of Kazakhstan, particularly in the context of the implementation gap between the standards of digital competence and the current state of the education system. The research is based on the application of two main methodological models for the evaluation of the level of educator competence, in which the European Framework for the Digital Competence of Educators is considered the scientific basis for the analysis of the areas of professional growth, while the TPACK model is applied to evaluate the efficiency of the educational process in the context of the intersection of technology, pedagogy, and subject matter content.
The analysis of the empirical data of the TALIS-2024 study allows for the qualitative comparison of the current state of the education system in Kazakhstan with the global trends in the context of OECD countries. It is stated that the educators of Kazakhstan exhibit a high level of "Digital Optimism," in which the level of teacher confidence in the level of technological knowledge is recorded at 75.12 points, which is significantly higher than the OECD average of 70.1. However, it is important to note that, in spite of the high level of self-assessment, there is a significant need for professional growth, in which 46.59% of educators require the acquisition of basic ICT skills. The progress in the development of the education system is limited by the "infrastructure ceiling," in which there is a lack of digital resources in 22.83% of the schools and unstable internet access in 22.69% of the educational environment.
The research also further highlights how Kazakhstan’s most important asset in this process of change is its already existing culture of professionalism and mutual support.
It has also been recommended that, while systemic changes are necessary at a national level, teachers themselves need to take an active role in developing their own personal Digital Roadmap. This involves a systematic process of self-assessment through the DigCompEdu SELFIE tool, with a further need for teachers to develop their Technological Knowledge (TK) in order to more effectively integrate this into TPACK. In conclusion, it has been established how Kazakhstan already has a robust foundation of digital optimism, with a world-class culture of teacher support. By closing the already existing infrastructure gap, it is considered that the potential for digital excellence within schools throughout Kazakhstan is limitless.
Artificial intelligence (AI) and gamification are becoming important tools in modern education. Gamification uses elements such as points, rewards, and challenges to increase student motivation and engagement. However, its effectiveness depends on learners’ interests, abilities, and the quality of game design.
Personalized learning aims to adapt content and tasks to individual student needs, but teachers often struggle to do this in traditional classrooms due to limited time and large class sizes. AI can support personalization by analyzing student performance, participation, and learning difficulties. Based on this data, AI systems can recommend appropriate materials, adjust task difficulty, and create individual learning paths.
This article compares educational platforms in Germany and Kazakhstan. Germany is well prepared to integrate AI into education and widely uses AI-based personalization and gamification. In contrast, many Kazakhstani platforms rely mainly on video lessons with limited interactive features.
The article also presents the educational chatbot “Help YOU!” for schools in Kazakhstan. By combining AI and gamification, the chatbot provides personalized student support and reduces teachers’ workload, demonstrating the potential of these technologies to improve education.
This longitudinal research paper investigates digital transformation of the traditional education sector of Kazakhstan as it pertains to Edtech (educational technologies) and progressive use of artificial intelligence (AI) in classrooms by developing a strategic approach to the creation of an eco-system of education on a national level. The focus of the study is to investigate the barriers that need to be addressed to enable Kazakhstan to achieve its goal of transitioning to an eco-system of education. Research Problem: The current situation in Kazakhstan is that billions of dollars have been invested in the creation of a digital infrastructure to support the implementation of digital initiatives throughout the country. However, the country has been unable to bring its diverse digital initiatives together into a cohesive programme due to the fact that there is still a gap between the ability of the country to implement technology and the improvement of educational outcomes of its students. The primary hindrance to bringing together the successful use of all of the technologies being implemented is due to a lack of a comprehensive strategy to integrate and govern the use of technology in education. Methodology: To develop the recommendations for transitioning Kazakhstan to the creation of an eco-system of education, the researchers utilised a comparative analysis of three strategic models of Edtech governance from leading educational systems of Singapore, Finland, and Germany. The rationale for choosing these three countries is based on the differences in their methods of implementation in education.
In this thesis, I examine the relationship between internal governance and university performance within the context of Russian higher education from 2012 to 2020, a period marked by the prominent application of New Public Management (NPM) instruments. This study investigates several dimensions of internal governance and its connection to university performance. First, how do internal governance characteristics - such as centralization, stakeholder involvement, external communication, and strategic orientation - relate to university performance? Second, is there a relationship between institutional strategy adoption and university performance from the perspective of university department heads? Additionally, I explore the institutional structures and governance arrangements in Russian higher education, with particular attention to two interrelated developments: the adoption of NPM instruments and system stratification. The study draws on data from a national-level survey of university leaders and administrators, complemented by statistical information. Depending on the data structure and variable characteristics, various quantitative methods - from simple difference tests to conditional efficiency estimations - will be applied to address the research questions.
Identification and characterization of small molecules targeting the E. coli AcrAB-TolC efflux pump
(2026)
This dissertation focuses on the identification and characterization of efflux pump inhibitors targeting the main tripartite efflux pump in Escherichia coli, AcrAB-TolC. Tripartite efflux pumps are integral membrane complexes that confer antimicrobial resistance to Gram-negative bacteria by extruding antibiotics. Inhibiting efflux systems with small molecules represents a promising strategy for extending the spectrum of antibiotics, and restoring antibiotic susceptibility in multidrug-resistant bacteria. However, no efflux pump inhibitors have been approved for clinical use so far.
Two substances, LP-115 and carmofur, that represent a basis for the development of novel efflux pump inhibitors were discovered, while postulated AcrA inhibitors were shown to be non-specific binders. LP-115 was identified employing an in silico repurposing screen targeting the outer membrane factor TolC followed by microbiological validation and deconstruction of a hit compound into fragments. Binding to TolC and AcrB was confirmed using MST, and a ligand-induced destabilization of the efflux pump complex assembly was observed using dynamic light scattering. Cryo-EM provided detailed molecular insights into the binding site at the AcrA-TolC interface. Our results suggest that LP-115 is an efflux pump inhibitor with a novel mechanism of action that consists of disrupting the AcrAB-TolC efflux pump assembly. Carmofur was identified employing a microbiological repurposing screen focusing on antimicrobial potentiating effects, followed by microbiological and biophysical characterization of the interaction with the isolated efflux pump subunits using microscale thermophoresis, nano differential scanning fluorimetry, and dynamic light scattering. The synergistic activity of carmofur in combination with an AcrAB-TolC substrate was TolC-dependent and specific binding to TolC was observed. Thus, carmofur could be used as starting point for the development of novel efflux pump inhibitors.
Our modern society relies heavily on the availability and utilisation of rare earth elements and yttrium (REY) for high-tech products and processes, which provokes a growing release of these metals into the environment and draws attention to biological and ecotoxicological consequences of their increasing concentrations in the environment. However, research has long neglected the environmental behaviour of REY. Coupled with publications including incomplete REY sets or data of questionable analytical quality, many open questions remain.
This dissertation investigates samples from the biosphere and from the hydrosphere to shed light on the REY transfer at their interface. Duckweeds, widely occurring small water plants, and Norwegian fjord waters together with Baltic Sea outflow samples were chosen as main study objects from the biosphere and the hydrosphere, respectively.
The findings of the biosphere-focused part improve the characterisation of the duckweed reference material BCR-670 (Lemna minor) and highlight the necessity of comparable sample processing for validation of data quality.
All naturally grown duckweeds investigated are REY quasi-hyperaccumulators and share similarly shaped, mildly fractionated shale-normalised REY patterns without positive anthropogenic Gd anomalies, regardless of whether they grew in waters with or without anomalous Gd enrichment.
The hydrosphere-focused part presents the first evidence for constant anthropogenic Gd input into the Baltic Sea outflow. The data combined with literature data further suggest that this signal is transported to southern Norway. In future, it may reach fjord waters further north along the Norwegian coast.
Overall, this dissertation provides important new information about the fate of geogenic and anthropogenic REY at the hydrosphere-biosphere interface and highlights the relevance of basic research as the basis for understanding the complex REY transfer mechanisms across environmental compartments.
Climate change driven by anthropogenic CO₂ emissions requires effective mitigation strategies. Negative emission technologies (NETs), particularly ocean alkalinity enhancement (OAE), are promising because they increase ocean alkalinity and promote CO₂ sequestration. This dissertation examines how marine molecular biology can help assess ecological risks and the overall efficacy of OAE. It presents two risk assessments on bacterial community responses to alkalinity exposure and develops a framework for a novel biological proxy for monitoring, reporting, and verification (MRV) in OAE.
Chapter 1 provides a general introduction.
Chapter 2 investigates how gradually increased alkalinity affects pelagic bacterial communities using a mesocosm experiment with 16S rRNA gene sequencing and flow cytometry. Results show high structural resilience, but quantitative shifts in bacterial abundance linked to phytoplankton dynamics indicate indirect ecological effects of OAE.
Chapter 3 expands this work by comparing two OAE strategies: olivine dissolution and direct dissolved alkalinity addition. A mesocosm experiment assessed microbial responses in seawater and oyster gills (Ostrea edulis). Olivine increased pollution-tolerant and biofilm-forming taxa, while dissolved alkalinity caused minimal change. These findings suggest that dissolved alkalinity below 500 µmol L⁻¹ is a relatively safe OAE approach.
Chapter 4 proposes carbonic anhydrase (CA), a key enzyme in marine carbon cycling, as a biological proxy for evaluating OAE performance. Structured hypotheses outline how CA expression and activity assays could support future OAE MRV systems. The chapter recommends shifting resources from broad bacterial community assessments toward investigating how alkalization affects CA.
Mapping challenging confined underwater environments pushes the boundaries of what is possible for state-of-the-art robotics. Current state-of-the-art high-performance equipment allows already for accurate mapping in such scenarios. However, these systems are often expensive. Affordable underwater robotic systems and sensors come with significantly reduced capabilities. Especially sonars are necessary for mapping unknown environments, due to cluttered water resulting in bad visibility for vision based sensors. Yet, affordable sonar sensors suffer from higher noise levels, reduced accuracy, and limited coverage. Consequently, developing methods to achieve reliable and accurate mapping of challenging environments with affordable hardware remains an open research question.
This thesis presents a Fourier-SOFT in 2D (FS2D) registration method for robust matching of high-noise 2D sonar scans. A Simultaneous Localization and Mapping (SLAM) framework designed to the unique challenges of affordable Mechanical Scanning Sonars (MSS) is presented, integrating this FS2D registration method.
In the context of the digitization of cultural heritage, the Bunker Valentin Memorial in Bremen is surveyed, and maps of its multiple basins are generated.
Additionally, this thesis contributes an open dataset with accurate ground truth for development and benchmarking 2D sonar navigation, mapping, and SLAM algorithms.
Overall, this thesis demonstrates that, when the unique characteristics of affordable hard ware are considered correctly, and the methods are designed accordingly, affordable underwater robots can effectively map and explore challenging, unknown environments.
The BlueAUV design, the FS2D registration method, SLAM framework for affordable hardware, and an openly available dataset provide a foundation for advancing robust mapping of challenging underwater environments within the research community.
Industrial Demand Response (IDR) systems have emerged as a key enabler for enhancing grid flexibility, particularly as industries face increasing pressure to optimize energy consumption and integrate with renewable energy sources. However, despite their potential, the adoption and scalability of IDR solutions are limited by a range of technical, infrastructural, and organizational challenges. This dissertation investigates how emerging digital technologies—namely, Artificial Intelligence/Machine Learning (AI/ML) and the computing continuum (edge, fog, and cloud computing)—can be leveraged to overcome these limitations and enable scalable, intelligent, and interoperable IDR architectures.
The study addresses three core research questions. First, it develops a taxonomy of barriers to IDR adoption, distinguishing between technological and non-technological constraints. Second, it explores how distributed computing paradigms can mitigate these challenges by enabling real-time, privacy-aware, and latency-sensitive decision-making across industrial sites. Third, it examines the synergistic integration of AI/ML within the computing continuum, emphasizing methods such as federated learning, transfer learning, and multi-agent reinforcement learning to overcome issues related to data sparsity, system complexity, and semantic heterogeneity.
A reference architecture for IDR aggregators is proposed, combining layered intelligence, semantic interoperability, and orchestration mechanisms. This architecture is mapped to real-world cloud and open-source platforms to demonstrate its practical applicability. The findings confirm that the integration of AI/ML and distributed computing is not only feasible but essential for advancing the resilience, autonomy, and responsiveness of future industrial energy systems.
The rapid growth of large-scale software systems has led to the adoption of automatic error reporting platforms collecting millions of crash reports from real users. Central to these reports is the stack trace — a record of function calls leading to failure — which serves as a crucial diagnostic resource. However, the sheer volume, diversity, and redundancy of reports create bottlenecks: developers are overwhelmed by duplicates and highly variable submissions from the same defect, impeding efficient issue resolution.
Existing deduplication and triage solutions in industry and academia mainly rely on string-matching, information retrieval, or graph-based heuristics. While efficient, string and IR methods often miss semantic and contextual nuances; graph-based models lose detail about individual reports, reducing accuracy. These limitations cause missed linkages between related errors and fragmentation of bug databases. The lack of scalable algorithms, real-world benchmarks, and advanced learning methods further restricts current tools.
This dissertation advances automation of error report processing via stack trace analysis. It introduces (1) hybrid similarity metrics extending traditional techniques, (2) deep learning models for robust similarity estimation, (3) aggregation strategies leveraging group-level information, (4) scalable solutions for industrial use, (5) the first models for automated developer assignment in stack trace–centered triage, and (6) methods for interpreting and highlighting the most informative stack frames. The research is validated on multiple proprietary and open datasets, including new benchmarks released as part of this work. Together, these contributions provide a unified, reproducible foundation for scalable, accurate, and actionable error report deduplication, grouping, assignment, and tooling in real-world software engineering.
This dissertation presents the synthesis of novel reduced POMs functionalized with dimethylarsinate groups. Chapter 1 introduces POMs, emphasizing polyoxomolybdates. Chapter 2 reviews organofunctionalized POMs and the rationale for this study, building on prior work with dimethylarsinate-functionalized molybdenum POMs. Chapter 3 covers experimental methods, characterization techniques, and synthesis of heterophosphonic and arsonic acid ligands. Chapter 4 describes the synthesis and characterization of eight dimethylarsinate-functionalized phosphomolybdates(V), [RPMoV6O15(OH)3{AsO2(CH3)2}3]2− (R = H, HO, CH3, HO2CCH2, HO2CC2H4, C6H5, 4-FC6H4, 4-F3COC6H4), the monoanionic mixed-valent heptamolybdate [HOMoVIMoV6O15(OH)3{AsO2(CH3)2}3]−, and d-block metal-substituted analogues [MPMoV6O15(OH)3{AsO2(CH3)2}3]2− (M = Fe2+, Ni2+, Mn2+). Chapter 5 focuses on the synthesis, structural features, and antibacterial properties of eleven dimethylarsinate-functionalized arsenomolybdates(V), [RAsMoV6O15(OH)3{AsO2(CH3)2}3]2− (R = HO, CH3, C2H5, C6H5, 3,5-(HOOC)2C6H3, 4-FC6H4, 4-F3CC6H4, 4-F3COC6H4, 4-BrC6H4 and 4-N3C6H4) and [AsIIIMoV6O15(OH)3{AsO2(CH3)2}3]3−. All compounds were synthesized in aqueous media and characterized in the solid state by single-crystal X-ray diffraction, TGA, elemental analysis, FT-IR, and PXRD, while their stability in solution and the gas phase was probed using multinuclear NMR (1H, 31P, 19F, 13C), ESI-MS, ion mobility MS, and MS/MS. Chapter 6 describes the synthesis and characterization of a novel heterometallic, dimethylarsinate-capped wheel-type POM, MoV12WVI18O84{AsO2(CH3)2}18]18− (Mo12W18), prepared under mildly acidic aqueous conditions, with alternating MoV2 and WVI3 units forming a ring with a ~1.5 nm central cavity. Its solid-state and solution behavior were probed using single-crystal XRD, IR, TGA, elemental analysis, MAS and CPMAS NMR, multinuclear solution NMR (1H, 13C, 183W, DOSY), UV-Vis, Raman spectroscopy, and SAXS.
Digitalization and Lean Management as Tools for Increasing Efficiency in the Transport Industry
(2025)
The essay explores the integration of digitalization and Lean management in the transport industry, emphasizing their combined role in enhancing efficiency, flexibility, and sustainability. It outlines the origins of Lean management in the Toyota Production System and its adaptation to logistics and transport through practices such as 5S, Kaizen, and Just-in-Time. The paper highlights successful examples from global companies like DHL, UPS, Delta Airlines, and DB Schenker, demonstrating measurable improvements in productivity and cost reduction.
Digital technologies—including the Internet of Things, Artificial Intelligence, Big Data, and digital twins—strengthen Lean principles by enabling real-time data analysis, automation, and predictive decision-making. The essay also examines Germany’s leadership in transport digitalization and describes practical observations from the Mercedes-Benz plant in Bremen.
Finally, it discusses challenges such as cybersecurity, integration complexity, and ethical concerns, concluding that the synergy between Lean and digitalization forms the foundation for the future of transport—making it smarter, greener, and more resilient in the global economy.
This thesis investigates consistency challenges in distributed stream processing systems. Prior work on this topic has made significant progress, with many ideas being implemented in state-of-the-art Stream Processing Engines (SPEs). In this thesis, we focus on formal modeling to better characterize existing problems and explore potential improvements.
We introduce a formal model of delivery guarantees and show that deterministic SPEs can theoretically achieve lower latency than non-deterministic ones for exactly-once guarantee. This is supported by experimental results demonstrating that a novel deterministic implementation performs better than current alternatives.
The thesis also presents a formal model for substream management, identifying a lower bound on the additional network traffic required for detecting substream termination. A corresponding framework is implemented that meets this bound and demonstrates improved performance over existing approaches.
These results contribute formal foundations and practical techniques for improving the performance and predictability of distributed stream processing systems.
Rapidly expanding anthropogenic activities are generating increasing volumes of wastewater globally each year, the majority of which remains inadequately treated before being released into aquatic ecosystems. Microalgal technologies offer a promising alternative for nutrient recovery in wastewater treatment, demonstrating significant advantages over conventional methods that are often energy-intensive and costly. Inadequate treatment not only leads to environmental pollution but also results in the irreversible loss of valuable nutrients, thereby disrupting the nutrient cycle. In recent years, the extraction of bioactive compounds from microalgae has attracted substantial attention. However, much of the research has remained confined to laboratory-scale studies with a focus on either energy efficiency or bioproduct synthesis, limiting their practical applicability. A major bottleneck in the scalability of algal-based systems is the energy- and cost-intensive nature of biomass harvesting, which can contribute up to 20–30% of total downstream processing costs. Additionally, the dependence on sunlight and large land areas further restricts the feasibility of microalgae-based wastewater treatment technologies in diverse environments. This study addresses three critical challenges associated with algae-based wastewater treatment. First, an innovative cultivation approach was developed to enable continuous wastewater treatment across two contrasting seasonal conditions—summer and winter. Second, the characteristics of wastewater post-treatment were analysed to identify fouling factors affecting the harvesting process. Third, a novel strategy was implemented to induce “hyper compensation” and “luxury uptake” of inorganic phosphorus by microalgae, achieving an exceptional phosphorus recovery rate of nearly 96%. To fully capitalize on the treated biomass, a novel bioplastic/bio-composite was developed by combining polylactic acid with phosphorus-enriched microalgae.
The relationship between network structure (structural connectivity, SC) and network representations of dynamics (functional connectivity, FC) is a topic of high scientific interest both for advancing theoretical understanding of complex systems and for its relevance to a wide range of applications.
In this thesis, correlations between structural and functional connectivity were investigated distinguishing between synchronous and sequential activity of the nodes. The primary analysis encompasses applying different dynamical models to network architectures to explore how SC/FC correlations are shaped by variations in network topology, coupling strength, and intrinsic system parameters across excitable, chaotic, and oscillatory dynamics. A more detailed investigation was conducted on regular graphs of coupled logistic maps. Symbolic encoding of the initial dynamics was used to construct equivalent cellular automaton models, followed by an analysis of the structure of their resulting attractors. The influence of noise on SC/FC correlations was also explored. Finally, SC and the two types of FC were conceptualized in a hydrological case study. Structural and functional networks were constructed from data collected in the Walnut Gulch Experimental Watershed (Arizona, USA). SC/FC correlations served as metrics to describe event-level hydrological responses of the watershed after various rainfall events, and their relationships to hydrological quantities were analyzed.
Succinic acid (SA) is one of the most promising bio-based platform chemicals, as it can serve as a precursor in the synthesis of many industrially relevant chemicals and polymers. Yeasts are particularly desirable microbial hosts for SA production because they tolerate low environmental pH, allowing the direct production of the undissociated form of the acid, which results in a significantly simpler and more cost-effective overall process. The fermentation performance of yeast cell factories heavily depends on the activity of membrane transporters responsible for substrate import, metabolite exchange between intracellular compartments, and SA export from the cell. As such, membrane transporters represent key targets in metabolic engineering efforts and are the primary focus of this work. The first part of this thesis provides new insights into plasma membrane transporters from the acetate uptake transporter (AceTr) family, namely Ato1 and SatP, by identifying amino acid residues critical for their specificity and activity. This includes the engineering and characterisation of Ato1 and SatP variants capable of transporting SA in S. cerevisiae. The impact of these transporter variants on extracellular SA accumulation is assessed and compared with that of Dct-02, a known fungal SA exporter, revealing that AceTr homologues and Dct-02 have opposing effects on SA production in S. cerevisiae under industrially relevant conditions. The second part of this thesis optimizes SA production from glycerol via the CO₂-fixing reductive TCA pathway in S. cerevisiae. To increase carbon flux through this cytosolic pathway, mitochondrial membrane transporters are identified as particularly attractive engineering targets. Overall, this work expands the understanding of membrane transporters relevant for SA production and improves the fermentation performance of existing SA-producing yeast cell factories.
The innate immune response is the first line of defense against viral infection. Host pattern recognition receptors (PRRs) detect pathogen-associated molecular patterns (PAMPs) and trigger several signaling pathways following viral infections. One such pathway, the cGAS-STING pathway, detects cytosolic DNA to induce the production and secretion of type I interferons.
The cGAS-STING pathway is activated by the presence of cytosolic DNA. cGAS (cGMP-AMP synthase), a cytosolic DNA detector, binds double-stranded DNA, dimerizes, and catalyzes the production of the second messenger cyclic GMP-AMP (cGAMP) from GTP and ATP. STING, an ER adaptor protein, binds to cGAMP and becomes activated through dimerization, which results in its trafficking from the ER to the Golgi. At the Golgi, STING recruits the kinase TBK1 and the transcription factor IRF3. TBK1 phosphorylates STING, itself, and IRF3. Phosphorylation activates IRF3, causing its translocation to the nucleus, where it induces the production of type I interferons.
To counteract the host immune response, human cytomegalovirus (HCMV) encodes several immunoevasin proteins. HCMV glycoproteins such as US6 inhibit antigen presentation by blocking peptide transport via the transporter associated with antigen processing (TAP). The region spanning amino acids 89–108 of US6 was identified as responsible for TAP inhibition.
Our studies have identified a novel interaction and function of US6. For the first time, we show that US6 interacts with the host p24 proteins TMED2 and TMED10. US6 inhibits the cytosolic DNA-triggered cGAS-STING pathway and reduces the production of IFNβ1. We have discovered a correlation between the binding of US6 to TMED2 and TMED10 and its ability to inhibit the production of type I interferons. Using sequential mutants, we show that US6 possesses two distinct and separable regions responsible for its functions. Microscopy reveals that US6 delays STING trafficking from the ER to the Golgi.
The theory of non-interacting Bose gases is supplemented by a numerical quantum field description with a two-dimensional non-local order parameter that allows the modeling of wave-like atomic correlations and interference effects in the limit of low atomic densities. From the present model, it is possible to explain symmetry aspects of non-interacting and very weakly interacting Bose gases in the limit of fluctuating particle numbers, like the forward propagation of time and the relation to the breaking and preservation of phase gauge symmetry in solids. In the present formalism, the propagation of one-directional time arises from the pre-defined and equivalent convergence of independent quantum fields towards the Boltzmann equilibrium, and it is shown that Glauber coherent states are related to the definition of the quantized field. Coherently coupling condensate and non-condensate parts as a direct consequence of the increasing quantum coherence time between the different quantum field components in the Bose gas from cooling to below the critical temperature, the present model describes symmetry breaking, which is originally known from the definition of a specific gauge field from Elitzur’s theorem for local gauge fields, as a global physical rather than a purely formal mathematical process.
The transition toward sustainable urban mobility and industrial efficiency requires decision-making tools that go beyond correlation-based analysis to uncover true cause-and-effect relationships. Traditional machine learning models, while effective for prediction, often act as "black boxes," lacking interpretability and failing to reveal the mechanisms underlying complex systems. To address these limitations, this dissertation introduces a modular Causal AI framework for smart decision-making, integrating causal discovery and inference with structured domain knowledge to enhance sustainability outcomes.
The framework is validated across three key domains: (1) urban CO2 emissions, (2) shared mobility demand, and (3) SME energy use. The first case study analyzes over 500,000 vehicles to uncover how engine performance and maintenance drive urban emissions. The second study examines shared bike systems, identifying causal impacts of weather patterns, station topology, and temporal demand fluctuations, supporting more adaptive fleet operations. The third applies the framework in a manufacturing SME, identifying the root causes of energy inefficiency and enabling targeted interventions to improve operational performance without compromising productivity.
This research advances the interpretability and actionability of AI in sustainability contexts by replacing opaque predictive models with transparent, evidence-based causal reasoning. Algorithms such as PC, FCI, GES, and DirectLiNGAM are employed alongside domain ontologies to uncover valid causal relationships and support decision-making. A hybrid approach also addresses feature selection, dimensionality reduction, and model explainability, making the methodology broadly applicable across diverse sustainability challenges.
While the framework demonstrates strong applicability, future work may focus on enhancing real-time scalability, adaptive ontology integration, and broader validation across domains such as electric mobility and smart energy systems. Overall, this thesis contributes a generalizable, interpretable Causal AI framework that enhances systemic understanding and supports sustainable transformation in policy, planning, and industrial decision-making.
As e-commerce continues to grow at a rapid pace, the ability to comprehend, segment, and predict customer behavior has become the core of business success. The objective of this project report is to study customer segmentation and predictive modeling using data mining methods on real-life online retail datasets. Exploiting Recency-Frequency-Monetary (RFM), K-Means Clustering, and Predictive Modeling (Logistic Regression, RandomForest, XGBoost, Deep Learning), the investigation explores novel customer segments. It assesses model performance demanding high value from customers. Findings indicate useful implications for segment behavior with the Deep Learning model giving outstanding performance in this case (accuracy up to 87.4% and ROC AUC of 0.932). The RFM-KMeans segmentation approach exposes tactical marketing opportunities in different customer groups, such as Champions, At-Risk Customers, and Big Spenders. This paper proposes a pipeline for scalable and interpretable analyses that combine unsupervised/supervised learning methods to support targeted marketing, retention forecasting, and long-tail customer value maximization in digital retail environments.
Energy management and sustainability have become critical global priorities in response to growing environmental concerns and the need to optimize resource consumption. As industries expand and technological advancements continue to shape modern societies, energy demands are rapidly rising. This leads to escalating levels of carbon emissions and resource depletion, negatively impacting the environment and the economy.
The thesis proposes an innovative approach to addressing energy management and sustainability complexities. The research focuses on developing a framework for smart energy systems that can autonomously improve their performance through knowledge sharing and semantic interoperability. The core idea behind this thesis is to enable smart energy systems to self-develop their knowledge models through decentralized technologies, particularly blockchain while ensuring peer-to-peer semantic interaction and collaboration across different environments and supply chains.
A key innovation of this thesis is the use of blockchain technology as the underlying platform for achieving semantic interoperability and knowledge exchange among smart systems. By leveraging blockchain’s decentralized nature, a peer-to-peer semantic interaction framework is established. The research introduces a smart contract mechanism and a token-based economic model to incentivize stakeholders within the blockchain network, ensuring that participants align with the sustainability goals of the network.
The thesis presents a novel approach to storing and exchanging knowledge models on the blockchain using the InterPlanetary File System (IPFS). This enables real-time updates to smart systems' knowledge models, allowing them to adapt and respond dynamically to changing environmental conditions and data inputs. Through the proposed blockchain ecosystem, the research provides a comprehensive solution for enhancing the interoperability, autonomy, and sustainability of smart energy systems.
This thesis addresses the challenge of accurately representing oceanic dynamics characterized by a multitude of interacting processes in numerical models. Specifically, it focuses on the simulation of oceanic circular patterns ranging from 10 to 100 km in diameter, known as mesoscale eddies. These eddies play a critical role in transporting energy, water properties, and nutrients across the ocean. This research uses grid resolutions that directly capture some larger mesoscale eddies (resolved) while employing advanced mathematical techniques to represent the effects of smaller, unresolved eddies. The primary aim of the thesis is to develop and incorporate novel mathematical and numerical approaches into the Finite Volume Sea Ice-Ocean Model (FESOM2) to improve the representation of mesoscale eddies while maintaining manageable computational costs.
To bridge the gap between low-resolution and high-resolution simulations, the study enhances the mesoscale eddy modeling framework formulated by Juricke et al. (2019) through the implementation of new components that address unresolved dynamics. This includes the addition of an advection-based component to capture nonlinear interactions between resolved and unresolved eddies, which demonstrates positive performance. Furthermore, stochastic elements are introduced into the governing equations to better represent small-scale variability missing from deterministic formulations. In parallel, the thesis explores alternative and complementary parameterization strategies, offering fresh perspectives on modeling at partially resolved scales. Each enhancement is rigorously evaluated using a suite of diagnostic tools — many developed as part of this work — with a particular focus on spectral analysis and energy pathways. Overall, the thesis proposes an integrated approach to mesoscale eddy modeling, advancing the accuracy and consistency of ocean simulations across eddy-permitting resolutions.
This project investigates customer behavior and pricing dynamics in the context of online apparel sales, using a real-world clickstream from a European e-commerce platform. Through a
structured process of exploratory data analysis and predictive modeling, the study explores how product visuals, attributes, and browsing behavior influence both purchasing patterns and pricing outcomes. Random Forest models were used for both regression and classification tasks to predict product prices and classify items into budget or premium tiers. These model outputs were then combined to detect potential pricing–perception mismatches, where a product appears over- or underpriced relative to how it is perceived based on its features. This approach helps highlight products that may benefit from a review of their pricing or presentation strategy. Key insights from the analysis show that products with frontal model photography tend to perform better in both pricing and sales, while black and blue items generate the highest revenue. These findings lead to clear business recommendations in areas such as pricing, visual merchandising, and product positioning. The project demonstrates how machine learning can be applied not only to forecast outcomes but also to guide practical, data-driven decisions in e-commerce.
Environmental, Social, and Governance (ESG) reporting is increasingly critical for corporate transparency and accountability as demands from investors, regulators, and the public grow. Despite its importance, ESG reporting faces persistent challenges, including fragmented standards, inconsistent metrics, misalignment with global goals like the UN SDGs, and limited relevance for stakeholders. These issues weaken benchmarking, data reliability, and decision-making, raising the risk of greenwashing. Although previous studies have explored drivers of ESG performance, there is still a clear need for technical solutions that enhance the quality and usefulness of ESG disclosures through integrated approaches.
This dissertation addresses these critical gaps by developing and proposing a novel, integrated framework that leverages the strength of semantic technologies, MCDM methods, and ESG maturity models. The research is structured through several interconnected studies, exploring specific industry applications using multi-criteria analysis techniques to identify and prioritize relevant ESG KPIs. A systematic literature review (SLR) provides a foundational understanding of existing ontology-driven solutions and their limitations in addressing reporting challenges and integrating quantitative methods. Based on these insights, the core contribution is the design of an ontology-based framework called ESGOnt. This framework utilizes a modular ESG ontology to standardize terminology, integrate fragmented data sources, and explicitly map ESG metrics to SDG targets.
This research contributes by addressing critical challenges in ESG reporting through a novel, integrated framework. It advances the understanding of how semantic models can be combined with quantitative methods to create robust, transparent, and actionable sustainability reporting systems, strengthening corporate accountability and supporting more effective contributions to global sustainable development.
This study conducts an extensive sentiment analysis of 7,357 English Tesla-related tweets using an XGBoost classifier, addressing the critical need to understand public perception of innovative companies in the electric vehicle (EV) sector (Jain et al., 2019). The methodology involves advanced preprocessing with tweet-preprocessor and NLTK, feature engineering using TF-IDF (2,000 features) and weighted VADER sentiment scores, and model optimization via GridSearchCV with SMOTE balancing (Chawla et al., 2002). The model achieved an accuracy of 71.67% and a macro F1-score of 67.73% ± 5.97%, with a sentiment distribution of 37.31% negative, 30.58% neutral, and 32.11% positive. Theoretical assumptions explore the impact of social media on EV sentiment (Thelwall et al., 2010), while results and discussions highlight model performance and Tesla-specific insights (Chen & Guestrin, 2016). The study concludes with implications for EV marketing and future research directions in NLP.
This dissertation examines Free/Libre Open Source Software (FLOSS) development groups through three interconnected studies, each applying computational social science methods to understand different aspects of online collaboration. The first study analyzes group interactions via pull requests, identifying five organizational structures ranging from hierarchical to collaboratively governed networks. This typology moves beyond the traditional "bazaar-cathedral" dichotomy and reveals how group structure impacts outcomes such as popularity, stability, and productivity. The second study employs an agent-based model informed by Affect Control Theory to explore how cultural dynamics shape roles, status, power distribution, and gender biases within FLOSS communities. Findings illustrate the interplay between cultural norms and social structures, highlighting pathways toward gender equality through cultural norm shifts. The third study investigates macro-level project dynamics, examining how repository fitness, preferential attachment, and aging influence project popularity. It compares the meritocratic nature of scientific research and FLOSS development, employing generative probabilistic models based on stochastic processes to understand the mechanisms driving popularity. Together, these studies demonstrate how platform design, cultural norms, and social structures collectively shape FLOSS projects, advancing our understanding of digital collaborative systems.
Tests and proofs are two main techniques in modern software verification. To test a program means running the program to check if its execution yields an expected outcome. To prove a program is to build a mathematical proof, showing the correctness of the program against its desired properties. In the traditional view, however, tests and proofs are considered as two incompatible techniques. They are often treated as warring siblings and mostly applied in isolation. The complementarity of tests and proofs — though not immediately apparent — has been relatively underexplored. Can their combination mitigate each other’s weaknesses while harnessing their respective strengths? This thesis tries to straddle the border between tests and proofs and suggest a concrete answer. It explores how the two approaches can collaborate with and mutually benefit one another. Three key contributions arise from this exploration. The first contribution is Proof2Test, a framework that transforms failed proofs into useful test cases, allowing programmers to use tests to debug failed proofs effectively. The second contribution consists of several proof-based test generation strategies, which use proofs to enhance both the efficiency and effectiveness of test generation. This thesis also extends SC with “loop unrolling”, considering not just zero or one but any number of iterations, up to a set limit. It also includes an empirical study to examine how much (if anything) testing strategies miss when they limit themselves to standard branch coverage and, conversely, how many more bugs we can find if we unroll loops. The last contribution of this thesis is an automatic program repair approach, Proof2Fix, which takes advantages of the proposed test generation methods to produce meaningful corrections to faults revealed by proofs.
This Ph.D. dissertation investigates the neurobehavioral, psychological, and cognitive factors influencing postural control and its age-related changes through four studies. Postural control is essential for daily activities, and its decline with age increases fall risk, leading to autonomy loss and reduced quality of life.
The first study (Imani & Godde, 2021) explores how self-efficacy mediates the relationship between falls and autonomy in older adults. While falls negatively impact autonomy, higher self-efficacy reduces this effect. Cognitive function predicts autonomy but does not moderate the relationship between falls and autonomy.
The second study examines the bidirectional link between cognitive and physical function. Results show that cognitive decline affects physical abilities over time and vice versa, highlighting the strong connection between executive function and motor control. This reinforces how cognitive decline, falls, and depressive mood contribute to reduced social participation.
The third study investigates the effects of transcranial direct current stimulation (tDCS) combined with balance training in younger adults. Targeting the sensorimotor cortex and dorsolateral prefrontal cortex (DLPFC), the study assesses how anodal tDCS enhances balance. Resting-state EEG is also explored as a predictor of training effects.
The fourth study extends this research to older adults, revealing age-related differences in neural activation during balance training. Results suggest individualized stimulation protocols may improve balance outcomes.
Overall, these studies emphasize the cognitive-motor link in postural control and propose targeted interventions to enhance balance and autonomy, particularly in older adults at risk of falls.
Entrepreneurship is vital in driving innovation, economic growth, and societal transformation. Due to its growing relevance, this dissertation examines the factors that motivate individuals to engage in entrepreneurial endeavors, grounded in the Theory of Planned Behavior (TPB) and enriched by supplementary theoretical perspectives. The research combines qualitative and quantitative methods to explore how attitudes, subjective norms, and perceived behavioral control interact with these diverse aspects. The first study identifies key influences on entrepreneurial intention through qualitative interviews with entrepreneurs, thereby revealing ten critical determinants. The second study expands upon these findings by employing a quantitative approach to validate these parameters through a survey among entrepreneurs in Germany. It demonstrates the central importance of economics-oriented education, materialistic values, and resilience while refining the theoretical assumptions of the TPB in entrepreneurial contexts. Lastly, the third study extends the analysis by comparing the personality traits of entrepreneurs and employees utilizing the OCEAN model. This research delineates substantial differences across all five traits, underscoring the psychological dimensions of entrepreneurial behavior. By amalgamating the three studies and their respective findings, this dissertation contributes to the theoretical understanding of entrepreneurship by enhancing the predictive validity of the TPB for research on entrepreneurship and highlighting differences in personality traits among entrepreneurs and employees. Through the integration of qualitative insights and quantitative validation, this dissertation provides a comprehensive and nuanced perspective on the complex nature of entrepreneurship.
