Publication Server of Constructor University Library
Not a member yet
858 research outputs found
Sort by
Expanding the Scope of the k-Prototypes Algorithm - Addressing Issues in Cluster Analysis of Mixed-Type Data Arising from Real-World Applications
No full textCluster analysis is a common part of data analysis. Its aim is the identification of unknown structure in data and the determination of a partition with groups of objects as similar as possible (so-called clusters). In contrast to the frequent occurrence of mixed-type data in real-world applications, involving numerical as well as categorical features, research tends to concentrate on data containing exclusively numerical features. There are comparatively few methods for clustering mixed-type data, with the k-prototypes algorithm being presumably the most widely recognized. The purpose of this cumulative dissertation is to expand the scope of this clustering algorithm. It addresses aspects that are not treated in Huang's original publication of the k-prototype algorithm, including the validation of the number of clusters, variable selection of data to be clustered, imputation of incomplete data, algorithm initialization, and the integration of an alternative distance measure in the algorithm routine. These issues are covered as they are prevalent in the application of the k-prototypes algorithm on real-world data. In these clustering tasks, the user lacks knowledge about the optimal number of clusters or the most useful variables to determine the cluster partition. In addition, incomplete data often occur and need to be dealt with. The algorithm’s initialization is analyzed to optimize the iterative routine, which was originally published with a random-based choice of initial prototypes. Additionally, the distance-based partitioning algorithm is extended to ordinal data for distance calculation with the change of the algorithm’s distance measure. To conduct the research, simulation studies on artificially generated data are utilized as well as exemplary analyzes on real-world data
Physiology and genomics of new marine methane-oxidizing bacteria
No full textMethane is the most abundant hydrocarbon on Earth, and plays a vital role in the global carbon cycle. In marine ecosystems, large quantities of produced methane are oxidized by methane-oxidizing microorganisms before it reaches the atmosphere. Aerobic methanotrophic bacteria, which consume methane in the upper oxic layers of marine sediments and the water column, represent the final oceanic methane filter. To date, the majority of marine aerobic methanotrophs remains uncultivated, with currently only nine formally described cultures. This hinders our understanding of their physiology that ultimately controls their activity and affects the dispersal of these methanotrophs in nature. In order to fill this gap in knowledge, the present work was focused on the isolation and characterization of marine methanotrophs from the North Sea and the Western Scheldt estuary sediments. The obtained methanotrophic cultures were investigated in physiological tests, and their metabolism was reconstructed based on high quality genomes. The isolation of four new methanotrophic species affiliated to the genera Methyloprofundus and Methylomarinum allowed to determine specific ecophysiological preferences and key conserved and distinct features within these genera. The isolate of Methylomarinum sarcina B3 exhibited a sarcina-like cell organization, which has not been previously reported for any marine methanotroph. The unusual Embden-Meyerhof-Parnas pathway identified in the new Methyloprofundus spp. could potentially serve as an alternative to the canonical glycolytic route. Finally, the discovery of a new putative nitrate reductase in Methyloprofundus spp. could have important implications for the understanding of the diversity of bacterial nitrate reductases and anaerobic respiration. Altogether, this work has advanced the characterization of the Methyloprofundus and Methylomarinum genera and laid the foundation for future research in microbial carbon and nitrogen metabolism
Unveiling Small Microplastics from European Waters to the Arctic: Surface Water to Deep Sediment and Reflections on Data Representativeness
No full textSince the invention of the first synthetic polymer in 1907, plastics have revolutionized industries but have also caused significant environmental challenges. Over 170 trillion plastic particles are estimated to float in the world's oceans. Once in the marine environment, these plastics fragment into smaller particles (microplastics, MPs, <5 mm) under environmental forces, contaminating ecosystems worldwide, including the remote Arctic.
This thesis investigates the Norwegian Coastal Current (NCC), a key transport route for MPs from northern Europe to the Arctic Ocean. Using novel sampling devices and advanced micro-Fourier transform infrared microscopy (μFTIR), small MPs (SMPs, 11–300 μm) were analyzed in seawater and sediments, providing the first detailed assessment of their spatial and temporal distribution in the NCC.
The results reveal the prevalence of SMPs from surface seawater to deep sediments, including layers deposited before the advent of plastics. Key findings include a relatively homogeneous horizontal distribution of SMPs in surface and subsurface seawater and significant variability in sediment concentrations (54–12491 MP kg⁻¹) across cores. SMP accumulation trends in post-1950 sediment layers varied, challenging their reliability as markers of the Anthropocene. A total of 21 polymer types were identified, with smaller size classes dominating, highlighting their ecological significance.
Further analysis of data representativeness revealed significant variability in MP concentrations and polymer diversity across stations, emphasizing the need for standardized protocols to ensure reliable data.
Despite being based on a single research cruise, this study provides a valuable snapshot of SMP distribution in the NCC. The findings critically evaluate current MP research practices and highlight the need for robust methodologies to improve the reliability of future studies
Improving pregnant women´s safe communication by applying health psychology and digital interventions: Evaluation of synchronous and asynchronous intervention approaches
No full textPatient safety, as a topic that has been under constant development for several decades with the aim of increasing it, is associated with an understudied target behaviour: safe communication behaviour of pregnant women.
Study one refers to N= 424 cross-sectional self-administered data. The evaluation was carried out via path modelling.
In the second study (N=367), the effectiveness of two forms of intervention developed in the project for improving safe communication was tested using repeated measures analyses of covariance (ANCOVAs) for a before-and-after comparison, whereby the online live seminar and control group followed an RCT design.
In the third study (N=1187), psychological predictors of safe communication as well as sociodemographic characteristics were identified using hierarchical regression. Risk factors associated with early drop out within the web app were identified using logistic regression.
Results from Study 1 show that an adapted HAPA fitted the data best, whereby two sequential mediations emerged. Regarding Study 2 results indicate that women participating in the digital live seminar improved their safe communication behaviour and perceived patient more compared to women using the web app. Study 3 identified that younger women are at risk for early dropout in the web app. Action planning revealed as a core finding in predicting safe communication behaviour over the course of the web app.
Psychological mechanisms and their social-cognitive determinants in the motivational phase of pregnant women's intention to communicate safely are mediated by coping-specific as well as planning-specific volitional determinants and could be identified and explained. Younger age is a risk factor for early discontinuation in the web-app. This thesis shows how digital applications could be built, developed and implemented from a psychological perspective in the future and is addressed to pregnant women, to health care providers and to researchers
Microbial physiology of nitric oxide-transforming microorganisms
No full textNitric oxide (NO) is a small gaseous molecule with important functions in cell biology and atmospheric chemistry owed to its unique physical and chemical properties. Since its relevance in biology was established, research on NO has focused primarily on its roles as signaling molecule, cytotoxin, and metabolic intermediate. Indeed, as a free radical and highly reactive compound, NO is as a potent toxin that can inhibit microbial growth, however it also has a central position in the microbial nitrogen cycle as a key intermediate in processes such as denitrification, aerobic ammonia oxidation, anaerobic ammonium oxidation, and nitrite-dependent anaerobic methane oxidation. Additionally, NO is a very energy-rich molecule with a high redox potential (NO/N2O; E0’ = +1.175 V) and it may have played a key role in the evolution of life on early Earth and the bioenergetic pathways related to modern denitrification and aerobic respiration. During recent years, we have been presented with new roles of NO in the nitrogen cycle. It appears as if the focus of NO research has slowly started to change its course as we begin to recognize its potential as direct substrate for microbial growth. Given its important roles in past and present microbial life, we believe that there must be a plethora of microorganisms that are capable of growing on NO conversions. Therefore, the main goal of my PhD project was to challenge our understanding of NO as mere toxin and intermediate, and investigate its potential as direct energy source for microbial life, whether it is through known or novel biochemical reactions, and the microorganisms that use it for this purpose, using a combination of continuous and batch incubations, physiological experiments, and multi-OMIC analyses
The Complex Effects of Distorted Social Perceptions on Opinions about Climate Change
No full textPolarisation is a great concern in current social and political debates. A divergence of opinions or, more generally, a lack of societal agreement, for example on fundamental problems like climate change, presents a barrier to rapid action against a looming crisis. There are many theories on why people polarise on certain topics. However, the drivers of polarisation in social environments are multi-faceted and involve complex feedbacks among social, cognitive, and structural processes. While humans require interactions with each other to form shared views and cooperate effectively on many problems, social influence can produce a variety of opinion patterns, such as consensus, persistent disagreement, or polarisation. In this thesis, I develop mathematical models of opinion formation or perception to uncover the conditions underpinning the emergence of such patterns. I formalise how psychological factors distort the way individuals perceive others into a mathematical language and analyse how these perceptions affect the formation of consensus or the persistence of disagreement in a virtual society. The factors are: (1) noise, (2) bias, or (3) subjective perception. Taken together, the three studies demonstrate that these factors distorting people's perceptions or responses to social influence have a non-negligible and sometimes surprising impact on collective opinion patterns. This thesis highlights the importance to better understand the mechanisms behind social phenomena and their non-trivial consequences on opinion dynamics. While the models and the conclusions presented in the thesis may not be readily used to predict opinion patterns, owing to the complexity and inherent uncertainty of our society, they contribute to the social sciences by demonstrating counter-intuitive consequences of seemingly obvious theoretical assumptions, highlighting gaps and potentially critical ambiguities in social theories, and suggesting future directions for empirical analysis
Advancing quantitative methods for complex social-ecological system research: a case study of aquaculture
No full textOver the last few decades, environmental governance research has embraced a complex adaptive systems (CAS) framing: solving sustainability challenges requires an understanding of the social-ecological systems (SES) they are embedded in, consisting of interlinked components and relationships which form dynamic and emergent patterns. Many frameworks have been developed to help conceptually understand SES, however less focus has been given to advancing methods for SES research. I identify a particular need to advance quantitative SES methods, as despite a growing range of available approaches, much quantitative SES research heavily relies on classic statistical methods which by design ignore interactive effects and focus on reducing systems to individual variables. This creates tensions when applied to systems shaped by highly interactive and context-sensitive processes. Further, despite an emphasis on standardizability, quantitative research has not led to widespread synthesis of SES knowledge. There is a need to advance quantitative SES methods in ways which 1) incorporate complex system properties into case studies and 2) synthesize generalizable findings across cases without overly abstracting case complexity. In this thesis I explore these methodological challenges within the literature on Elinor Ostrom’s social-ecological systems framework (SESF). I then apply recent advances in methods for complexity through the case study of small-scale aquaculture governance in Indonesia: a participatory modeling method called fuzzy cognitive mapping to analyze “mental models” of aquaculture complexity, and archetypes analysis to synthesize generalizable patterns in complexity across a large set of heterogeneous aquaculture cases. I conclude that advancing SES methods to inform sustainable outcomes requires more critical engagement with “complex systems thinking” in not only conceptualizing environmental governance problems but also in empirical research design
Machine Learning-Based Scheduling in Steel Manufacturing
No full textSteel manufacturing is characterized by its high energy consumption and the production of high-value-added products. In real-world steel production, unforeseen events frequently disrupt schedules, emphasizing the critical need for effective and adaptive planning to ensure continuous operations. This study makes significant contributions to the steel industry by offering new approaches to improve efficiency, streamline operations, and optimize production processes, ultimately driving advancements in steel manufacturing performance.
To address the challenges inherent to EAF-based steelmaking, a seamless pipeline of algorithms has been developed. This pipeline works together to enhance the manufacturing planning process by providing an accurate chemical condensation of molten steel and classifying this outcome according to the most feasible steel grade, which can be obtained with a minimum of purification. Finally, the pipeline reschedules the planning process with the objective of maximizing machine utilization.
To achieve these goals, the steelmaking stage key performance indicators (KPIs) that have the most impact on the quality of the final product were first identified. In the next step, a novel prediction algorithm utilizing a multilayer feedforward neural network was developed to estimate key quality parameters. Finally, to enhance the pipeline's resilience to disruptions, a Genetic Algorithm (GA) is employed to mitigate the impact of scenarios where processing times for jobs vary and do not align with the established schedule. This is a prevalent disruption event that renders the base schedule infeasible. The proposed algorithm aims to minimize total completion and waiting times, thereby enhancing operational efficiency and minimizing manufacturing costs
Beyond Western Influences: Development and Reforms of Social Protection and Pension Schemes in China since 1978
No full textOver the past 40 years, a modern welfare state has emerged in the People’s Republic of China (PRC). Social protection has thereby extended from a selected group of privileged state employees and civil servants to most of the population. This dissertation argues that the significant extension of social protection, and of pension schemes in particular, is not only a consequence of domestic economic development and political reform; international influences also manifested themselves during the course of the development and reforms of the Chinese social protection system.
My key research question is: How have international influences impacted on Chinese social security and pension reform? While I do analyze the roles of “Western” influences and International Organizations, I particularly focus on one important gap in the literature – the scarcely studied role of “Eastern” or East Asian influences on Chinese social protection development.
Theoretically, this dissertation builds on the policy learning literature as well as the concept of East Asian welfare productivism to explain how international influences impacted on Chinese social protection expansion and reforms and to better understand welfare features the PRC shares with its East Asian neighbors. Regarding methodology, this dissertation employs several methods in the tradition of qualitative research: process tracing, document analysis, and discourse analysis. The core data of this research was collected via expert interviews during fieldwork in China. Moreover, archive data, official policy reports in Chinese, English and Japanese, and other primary and secondary data were utilized.
This cumulative dissertation consists of four interdependent articles focusing on the development and expansion of Chinese social protection under international influences, and especially on the two major reforms of Chinese pension schemes: the 1990s Urban Employee Pension Reforms, and the 2009 New Rural Pension Scheme
Development and Application of Compound Class-Specific Benchmark Data Sets for Differentiated Assessment of Docking and Scoring Algorithm Performance
No full textThis thesis focuses on the development and application of benchmark data sets for diverse compound classes and the differentiated assessment of docking and scoring algorithm performance using the curated sets. Various popular software, including AutoDock, AutoDock Vina, GOLD, MOE, FlexX and FITTED were assessed for two important types of compounds, which are summarized as follows.
In publication I, we investigated the fragment placement performance of molecular docking software AutoDock, AutoDock Vina, GOLD and FlexX. For this assessment we constructed LEADS-FRAG, a benchmark data set containing 93 high-quality protein-fragment complexes. GOLD with ChemPLP and AutoDock Vina performed best and generated near-native conformations (root mean square deviation <1.5 Å) for more than 50% of the data set considering the top-ranked docking pose. Taking into account all docking poses, the tested programs generated near-native conformations for up to 86% of the fragments. By rescoring with the GOLD scoring functions and PLIff, the number of near-native conformations increased up to 40% with respect to the top-rescored poses, showing that conventional small-molecule docking programs achieve a satisfactory fragment docking performance.
In manuscript 2, we examined covalently bound ligands and tested the efficiency of covalent docking options in the software programs AutoDock, GOLD, MOE and FITTED. We generated the LEADS-COV data set, containing 89 high-quality covalently bound protein-ligand complexes: 47 with a cysteine bound ligand and 42 with serine. For Cysteine GOLD with ChemPLP or ChemScore performed best and generated near-native conformations (root mean square deviation <1.5 Å) for more than 40% of the data set considering the top-ranked docking pose. Serine in comparison had better results, with over 65% top-ranked near native poses by GOLD with ChemPLP. Taking into account all generated poses values went up to over 65% for cysteine and over 80% for serine