Refine
Has Fulltext
- yes (159) (remove)
Year of publication
- 2022 (159) (remove)
Document Type
- Doctoral Thesis (159) (remove)
Is part of the Bibliography
- yes (159)
Keywords
- Klimawandel (5)
- climate change (5)
- Digitalisierung (3)
- Modellierung (3)
- Röntgenspektroskopie (3)
- modelling (3)
- Adipositas (2)
- Arbeitszufriedenheit (2)
- Bewegungsökologie (2)
- Bundeswehr (2)
Institute
- Institut für Biochemie und Biologie (27)
- Extern (26)
- Institut für Physik und Astronomie (20)
- Institut für Chemie (19)
- Institut für Geowissenschaften (17)
- Hasso-Plattner-Institut für Digital Engineering GmbH (14)
- Institut für Umweltwissenschaften und Geographie (7)
- Wirtschaftswissenschaften (7)
- Institut für Ernährungswissenschaft (5)
- Institut für Mathematik (5)
- Digital Engineering Fakultät (4)
- Fachgruppe Volkswirtschaftslehre (4)
- Institut für Romanistik (4)
- Department Erziehungswissenschaft (3)
- Department Linguistik (3)
- Fachgruppe Politik- & Verwaltungswissenschaft (3)
- Psycholinguistics and Neurolinguistics (3)
- Department Sport- und Gesundheitswissenschaften (2)
- Fachgruppe Betriebswirtschaftslehre (2)
- Fachgruppe Soziologie (2)
- Department Grundschulpädagogik (1)
- Historisches Institut (1)
- Institut für Anglistik und Amerikanistik (1)
- Institut für Informatik und Computational Science (1)
- Institut für Jüdische Theologie (1)
- Institut für Künste und Medien (1)
- Institut für Philosophie (1)
- Sozialwissenschaften (1)
- Öffentliches Recht (1)
The development of speaking competence is widely regarded as a central aspect of second language (L2) learning. It may be questioned, however, if the currently predominant ways of conceptualising the term fully satisfy the complexity of the construct: Although there is growing recognition that language primarily constitutes a tool for communication and participation in social life, as yet it is rare for conceptualisations of speaking competence to incorporate the ability to inter-act and co-construct meaning with co-participants. Accordingly, skills allowing for the successful accomplishment of interactional tasks (such as orderly speaker change, and resolving hearing and understanding trouble) also remain largely unrepresented in language teaching and assessment. As fostering the ability to successfully use the L2 within social interaction should arguably be a main objective of language teaching, it appears pertinent to broaden the construct of speaking competence by incorporating interactional competence (IC). Despite there being a growing research interest in the conceptualisation and development of (L2) IC, much of the materials and instruments required for its teaching and assessment, and thus for fostering a broader understanding of speaking competence in the L2 classroom, still await development. This book introduces an approach to the identification of candidate criterial features for the assessment of EFL learners’ L2 repair skills. Based on a corpus of video-recorded interaction between EFL learners, and following conversation-analytic and interactional-linguistic methodology as well as drawing on basic premises of research in the framework of Conversation Analysis for Second Language Acquisition, differences between (groups of) learners in terms of their L2 repair conduct are investigated through qualitative and inductive analyses. Candidate criterial features are derived from the analysis results. This book does not only contribute to the operationalisation of L2 IC (and of L2 repair skills in particular), but also lays groundwork for the construction of assessment scales and rubrics geared towards the evaluation of EFL learners’ L2 interactional skills.
Knowledge graphs are structured repositories of knowledge that store facts
about the general world or a particular domain in terms of entities and
their relationships. Owing to the heterogeneity of use cases that are served
by them, there arises a need for the automated construction of domain-
specific knowledge graphs from texts. While there have been many research
efforts towards open information extraction for automated knowledge graph
construction, these techniques do not perform well in domain-specific settings.
Furthermore, regardless of whether they are constructed automatically from
specific texts or based on real-world facts that are constantly evolving, all
knowledge graphs inherently suffer from incompleteness as well as errors in
the information they hold.
This thesis investigates the challenges encountered during knowledge graph
construction and proposes techniques for their curation (a.k.a. refinement)
including the correction of semantic ambiguities and the completion of missing
facts. Firstly, we leverage existing approaches for the automatic construction
of a knowledge graph in the art domain with open information extraction
techniques and analyse their limitations. In particular, we focus on the
challenging task of named entity recognition for artwork titles and show
empirical evidence of performance improvement with our proposed solution
for the generation of annotated training data.
Towards the curation of existing knowledge graphs, we identify the issue of
polysemous relations that represent different semantics based on the context.
Having concrete semantics for relations is important for downstream appli-
cations (e.g. question answering) that are supported by knowledge graphs.
Therefore, we define the novel task of finding fine-grained relation semantics
in knowledge graphs and propose FineGReS, a data-driven technique that
discovers potential sub-relations with fine-grained meaning from existing pol-
ysemous relations. We leverage knowledge representation learning methods
that generate low-dimensional vectors (or embeddings) for knowledge graphs
to capture their semantics and structure. The efficacy and utility of the
proposed technique are demonstrated by comparing it with several baselines
on the entity classification use case.
Further, we explore the semantic representations in knowledge graph embed-
ding models. In the past decade, these models have shown state-of-the-art
results for the task of link prediction in the context of knowledge graph comple-
tion. In view of the popularity and widespread application of the embedding
techniques not only for link prediction but also for different semantic tasks,
this thesis presents a critical analysis of the embeddings by quantitatively
measuring their semantic capabilities. We investigate and discuss the reasons
for the shortcomings of embeddings in terms of the characteristics of the
underlying knowledge graph datasets and the training techniques used by
popular models.
Following up on this, we propose ReasonKGE, a novel method for generating
semantically enriched knowledge graph embeddings by taking into account the
semantics of the facts that are encapsulated by an ontology accompanying the
knowledge graph. With a targeted, reasoning-based method for generating
negative samples during the training of the models, ReasonKGE is able to
not only enhance the link prediction performance, but also reduce the number
of semantically inconsistent predictions made by the resultant embeddings,
thus improving the quality of knowledge graphs.
Distances affect economic decision-making in numerous situations. The time at which we make a decision about future consumption has an impact on our consumption behavior. The spatial distance to employer, school or university impacts the place where we live and vice versa. The emotional closeness to other individuals influences our willingness to give money to them. This cumulative thesis aims to enrich the literature on the role of distance for economic decision-making. Thereby, each of my research projects sheds light on the impact of one kind of distance for efficient decision-making.
The Antarctic ice sheet is the largest freshwater reservoir worldwide. If it were to melt completely, global sea levels would rise by about 58 m. Calculation of projections of the Antarctic contribution to sea level rise under global warming conditions is an ongoing effort which
yields large ranges in predictions. Among the reasons for this are uncertainties related to the physics of ice sheet modeling. These
uncertainties include two processes that could lead to runaway ice retreat: the Marine Ice Sheet Instability (MISI), which causes rapid grounding line retreat on retrograde bedrock, and the Marine Ice Cliff Instability (MICI), in which tall ice cliffs become unstable and calve off, exposing even taller ice cliffs.
In my thesis, I investigated both marine instabilities (MISI and MICI) using the Parallel Ice Sheet Model (PISM), with a focus on MICI.
An important goal in biotechnology and (bio-) medical research is the isolation of single cells from a heterogeneous cell population. These specialised cells are of great interest for bioproduction, diagnostics, drug development, (cancer) therapy and research. To tackle emerging questions, an ever finer differentiation between target cells and non-target cells is required. This precise differentiation is a challenge for a growing number of available methods.
Since the physiological properties of the cells are closely linked to their morphology, it is beneficial to include their appearance in the sorting decision. For established methods, this represents a non addressable parameter, requiring new methods for the identification and isolation of target cells. Consequently, a variety of new flow-based methods have been developed and presented in recent years utilising 2D imaging data to identify target cells within a sample. As these methods aim for high throughput, the devices developed typically require highly complex fluid handling techniques, making them expensive while offering limited image quality.
In this work, a new continuous flow system for image-based cell sorting was developed that uses dielectrophoresis to precisely handle cells in a microchannel. Dielectrophoretic forces are exerted by inhomogeneous alternating electric fields on polarisable particles (here: cells). In the present system, the electric fields can be switched on and off precisely and quickly by a signal generator. In addition to the resulting simple and effective cell handling, the system is characterised by the outstanding quality of the image data generated and its compatibility with standard microscopes. These aspects result in low complexity, making it both affordable and user-friendly.
With the developed cell sorting system, cells could be sorted reliably and efficiently according to their cytosolic staining as well as morphological properties at different optical magnifications. The achieved purity of the target cell population was up to 95% and about 85% of the sorted cells could be recovered from the system. Good agreement was achieved between the results obtained and theoretical considerations. The achieved throughput of the system was up to 12,000 cells per hour. Cell viability studies indicated a high biocompatibility of the system.
The results presented demonstrate the potential of image-based cell sorting using dielectrophoresis. The outstanding image quality and highly precise yet gentle handling of the cells set the system apart from other technologies. This results in enormous potential for processing valuable and sensitive cell samples.
Accurately solving classification problems nowadays is likely to be the most relevant machine learning task. Binary classification separating two classes only is algorithmically simpler but has fewer potential applications as many real-world problems are multi-class. On the reverse, separating only a subset of classes simplifies the classification task. Even though existing multi-class machine learning algorithms are very flexible regarding the number of classes, they assume that the target set Y is fixed and cannot be restricted once the training is finished. On the other hand, existing state-of-the-art production environments are becoming increasingly interconnected with the advance of Industry 4.0 and related technologies such that additional information can simplify the respective classification problems. In light of this, the main aim of this thesis is to introduce dynamic classification that generalizes multi-class classification such that the target class set can be restricted arbitrarily to a non-empty class subset M of Y at any time between two consecutive predictions.
This task is solved by a combination of two algorithmic approaches. First, classifier calibration, which transforms predictions into posterior probability estimates that are intended to be well calibrated. The analysis provided focuses on monotonic calibration and in particular corrects wrong statements that appeared in the literature. It also reveals that bin-based evaluation metrics, which became popular in recent years, are unjustified and should not be used at all. Next, the validity of Platt scaling, which is the most relevant parametric calibration approach, is analyzed in depth. In particular, its optimality for classifier predictions distributed according to four different families of probability distributions as well its equivalence with Beta calibration up to a sigmoidal preprocessing are proven. For non-monotonic calibration, extended variants on kernel density estimation and the ensemble method EKDE are introduced. Finally, the calibration techniques are evaluated using a simulation study with complete information as well as on a selection of 46 real-world data sets.
Building on this, classifier calibration is applied as part of decomposition-based classification that aims to reduce multi-class problems to simpler (usually binary) prediction tasks. For the involved fusing step performed at prediction time, a new approach based on evidence theory is presented that uses classifier calibration to model mass functions. This allows the analysis of decomposition-based classification against a strictly formal background and to prove closed-form equations for the overall combinations. Furthermore, the same formalism leads to a consistent integration of dynamic class information, yielding a theoretically justified and computationally tractable dynamic classification model. The insights gained from this modeling are combined with pairwise coupling, which is one of the most relevant reduction-based classification approaches, such that all individual predictions are combined with a weight. This not only generalizes existing works on pairwise coupling but also enables the integration of dynamic class information.
Lastly, a thorough empirical study is performed that compares all newly introduced approaches to existing state-of-the-art techniques. For this, evaluation metrics for dynamic classification are introduced that depend on corresponding sampling strategies. Thereafter, these are applied during a three-part evaluation. First, support vector machines and random forests are applied on 26 data sets from the UCI Machine Learning Repository. Second, two state-of-the-art deep neural networks are evaluated on five benchmark data sets from a relatively recent reference work. Here, computationally feasible strategies to apply the presented algorithms in combination with large-scale models are particularly relevant because a naive application is computationally intractable. Finally, reference data from a real-world process allowing the inclusion of dynamic class information are collected and evaluated. The results show that in combination with support vector machines and random forests, pairwise coupling approaches yield the best results, while in combination with deep neural networks, differences between the different approaches are mostly small to negligible. Most importantly, all results empirically confirm that dynamic classification succeeds in improving the respective prediction accuracies. Therefore, it is crucial to pass dynamic class information in respective applications, which requires an appropriate digital infrastructure.
The present dissertation conducts empirical research on the relationship between urban life and its economic costs, especially for the environment. On the one hand, existing gaps in research on the influence of population density on air quality are closed and, on the other hand, innovative policy measures in the transport sector are examined that are intended to make metropolitan areas more sustainable. The focus is on air pollution, congestion and traffic accidents, which are important for general welfare issues and represent significant cost factors for urban life. They affect a significant proportion of the world's population. While 55% of the world's people already lived in cities in 2018, this share is expected to reach approximately 68% by 2050.
The four self-contained chapters of this thesis can be divided into two sections: Chapters 2 and 3 provide new causal insights into the complex interplay between urban structures and air pollution. Chapters 4 and 5 then examine policy measures to promote non-motorised transport and their influence on air quality as well as congestion and traffic accidents.
Neural conversation models aim to predict appropriate contributions to a (given) conversation by using neural networks trained on dialogue data. A specific strand focuses on non-goal driven dialogues, first proposed by Ritter et al. (2011): They investigated the task of transforming an utterance into an appropriate reply. Then, this strand evolved into dialogue system approaches using long dialogue histories and additional background context. Contributing meaningful and appropriate to a conversation is a complex task, and therefore research in this area has been very diverse: Serban et al. (2016), for example, looked into utilizing variable length dialogue histories, Zhang et al. (2018) added additional context to the dialogue history, Wolf et al. (2019) proposed a model based on pre-trained Self-Attention neural networks (Vasvani et al., 2017), and Dinan et al. (2021) investigated safety issues of these approaches. This trend can be seen as a transformation from trying to somehow carry on a conversation to generating appropriate replies in a controlled and reliable way.
In this thesis, we first elaborate the meaning of appropriateness in the context of neural conversation models by drawing inspiration from the Cooperative Principle (Grice, 1975). We first define what an appropriate contribution has to be by operationalizing these maxims as demands on conversation models: being fluent, informative, consistent towards given context, coherent and following a social norm. Then, we identify different targets (or intervention points) to achieve the conversational appropriateness by investigating recent research in that field.
In this thesis, we investigate the aspect of consistency towards context in greater detail, being one aspect of our interpretation of appropriateness.
During the research, we developed a new context-based dialogue dataset (KOMODIS) that combines factual and opinionated context to dialogues. The KOMODIS
dataset is publicly available and we use the data in this thesis to gather new insights in context-augmented dialogue generation.
We further introduced a new way of encoding context within Self-Attention based neural networks. For that, we elaborate the issue of space complexity from knowledge graphs,
and propose a concise encoding strategy for structured context inspired from graph neural networks (Gilmer et al., 2017) to reduce the space complexity of the additional context. We discuss limitations of context-augmentation for neural conversation models, explore the characteristics of knowledge graphs, and explain how we create and augment knowledge graphs for our experiments.
Lastly, we analyzed the potential of reinforcement and transfer learning to improve context-consistency for neural conversation models. We find that current reward functions need to be more precise to enable the potential of reinforcement learning, and that sequential transfer learning can improve the subjective quality of generated dialogues.
This dissertation aimed to determine differential expressed miRNAs in the context of chronic pain in polyneuropathy. For this purpose, patients with chronic painful polyneuropathy were compared with age matched healthy patients. Taken together, all miRNA pre library preparation quality controls were successful and none of the samples was identified as an outlier or excluded for library preparation. Pre sequencing quality control showed that library preparation worked for all samples as well as that all samples were free of adapter dimers after BluePippin size selection and reached the minimum molarity for further processing. Thus, all samples were subjected to sequencing. The sequencing control parameters were in their optimal range and resulted in valid sequencing results with strong sample to sample correlation for all samples. The resulting FASTQ file of each miRNA library was analyzed and used to perform a differential expression analysis. The differentially expressed and filtered miRNAs were subjected to miRDB to perform a target prediction. Three of those four miRNAs were downregulated: hsa-miR-3135b, hsa-miR-584-5p and hsa-miR-12136, while one was upregulated: hsa-miR-550a-3p. miRNA target prediction showed that chronic pain in polyneuropathy might be the result of a combination of miRNA mediated high blood flow/pressure and neural activity dysregulations/disbalances. Thus, leading to the promising conclusion that these four miRNAs could serve as potential biomarkers for the diagnosis of chronic pain in polyneuropathy.
Since TRPV1 seems to be one of the major contributors of nociception and is associated with neuropathic pain, the influence of PKA phosphorylated ARMS on the sensitivity of TRPV1 as well as the part of AKAP79 during PKA phosphorylation of ARMS was characterized. Therefore, possible PKA-sites in the sequence of ARMS were identified. This revealed five canonical PKA-sites: S882, T903, S1251/52, S1439/40 and S1526/27. The single PKA-site mutants of ARMS revealed that PKA-mediated ARMS phosphorylation seems not to influence the interaction rate of TRPV1/ARMS. While phosphorylation of ARMST903 does not increase the interaction rate with TRPV1, ARMSS1526/27 is probably not phosphorylated and leads to an increased interaction rate. The calcium flux measurements indicated that the higher the interaction rate of TRPV1/ARMS, the lower the EC50 for capsaicin of TRPV1, independent of the PKA phosphorylation status of ARMS. In addition, the western blot analysis confirmed the previously observed TRPV1/ARMS interaction. More importantly, AKAP79 seems to be involved in the TRPV1/ARMS/PKA signaling complex. To overcome the problem of ARMS-mediated TRPV1 sensitization by interaction, ARMS was silenced by shRNA. ARMS silencing resulted in a restored TRPV1 desensitization without affecting the TRPV1 expression and therefore could be used as new topical therapeutic analgesic alternative to stop ARMS mediated TRPV1 sensitization.
In this thesis, I present my contributions to the field of ultrafast molecular spectroscopy. Using the molecule 2-thiouracil as an example, I use ultrashort x-ray pulses from free- electron lasers to study the relaxation dynamics of gas-phase molecular samples. Taking advantage of the x-ray typical element- and site-selectivity, I investigate the charge flow and geometrical changes in the excited states of 2-thiouracil.
In order to understand the photoinduced dynamics of molecules, knowledge about the ground-state structure and the relaxation after photoexcitation is crucial. Therefore, a part of this thesis covers the electronic ground-state spectroscopy of mainly 2-thiouracil to provide the basis for the time-resolved experiments. Many of the previously published studies that focused on the gas-phase time-resolved dynamics of thionated uracils after UV excitation relied on information from solution phase spectroscopy to determine the excitation energies. This is not an optimal strategy as solvents alter the absorption spec- trum and, hence, there is no guarantee that liquid-phase spectra resemble the gas-phase spectra. Therefore, I measured the UV-absorption spectra of all three thionated uracils to provide a gas-phase reference and, in combination with calculations, we determined the excited states involved in the transitions.
In contrast to the UV absorption, the literature on the x-ray spectroscopy of thionated uracil is sparse. Thus, we measured static photoelectron, Auger-Meitner and x-ray absorption spectra on the sulfur L edge before or parallel to the time-resolved experiments we performed at FLASH (DESY, Hamburg). In addition, (so far unpublished) measurements were performed at the synchrotron SOLEIL (France) which have been included in this thesis and show the spin-orbit splitting of the S 2p photoline and its satellite which was not observed at the free-electron laser.
The relaxation of 2-thiouracil has been studied extensively in recent years with ultrafast visible and ultraviolet methods showing the ultrafast nature of the molecular process after photoexcitation. Ultrafast spectroscopy probing the core-level electrons provides a complementary approach to common optical ultrafast techniques. The method inherits its local sensitivity from the strongly localised core electrons. The core energies and core-valence transitions are strongly affected by local valence charge and geometry changes, and past studies have utilised this sensitivity to investigate the molecular process reflected by the ultrafast dynamics. We have built an apparatus that provides the requirements to perform time-resolved x-ray spectroscopy on molecules in the gas phase. With the apparatus, we performed UV-pump x-ray-probe electron spectroscopy on the S 2p edge of 2-thiouracil using the free-electron laser FLASH2. While the UV triggers the relaxation dynamics, the x-ray probes the single sulfur atom inside the molecule. I implemented photoline self-referencing for the photoelectron spectral analysis. This minimises the spectral jitter of the FEL, which is due to the underlying self-amplified spontaneous emission (SASE) process. With this approach, we were not only able to study dynamical changes in the binding energy of the electrons but also to detect an oscillatory behaviour in the shift of the observed photoline, which we associate with non-adiabatic dynamics involving several electronic states. Moreover, we were able to link the UV-induced shift in binding energy to the local charge flow at the sulfur which is directly connected to the electronic state. Furthermore, the analysis of the Auger-Meitner electrons shows that energy shifts observed at early stages of the photoinduced relaxation are related to the geometry change in the molecule. More specifically, the observed increase in kinetic energy of the Auger-Meitner electrons correlates with a previously predicted C=S bond stretch.