Doctoral Theses / Doktorsavhandlingar Institutionen för filosofi, lingvistik och vetenskapsteori

Permanent URI for this collectionhttps://gupea-staging.ub.gu.se/handle/2077/19525

Browse

Now showing 1 - 20 of 55

Ordering Thought: Cognitive Complexity of a Description Logic
(2025-10-06) Fokkens, Tjeerd
\noindent Knowledge bases are used in a variety of industries to efficiently represent data, but they sometimes contain errors. The debugging of knowledge bases is, therefore, a vitally important process. Debugging requires explanations of entailments in description logic, because the latter are used for answering knowledge base queries. Such explanations can be generated by axiom pinpointing, which is a technique to find justifications: minimal sets of axioms that entail a certain conclusion. Typically, a large set of justifications is found. It is difficult to select the most explanatory one, i.e. the one which requires the least cognitive effort to understand. Four contributions are made with this thesis to solve this problem. First, the concept of relative cognitive complexity is defined. Second, the model SHARP is created with the cognitive architecture ACT-R, simulating the process of a human deciding the consistency of so-called ABoxes, the definition of which some justifications satisfy. The scope is restricted to ABoxes in the description logic $\mathcal{ALE}$. SHARP can be used to model cognitive effort, so as to capture the relative cognitive complexity of $\mathcal{ALE}$ ABoxes. Third, an experiment is performed to test the predictions on cognitive behaviour based on SHARP's simulation results. The model performs quite well on the relative cognitive complexity, but also shows some peculiar effects. Fourth, three surrogate modelling techniques were tested to decrease computation times: Random Forests (RF), Support Vector Regression (SVR) and Symbolic Regression (SR). The three techniques achieve similar performance, but SR achieves the lowest computation times, although it required a long training time.
Fixed IDs about Truth – Truth and Fixpoints over Intuitionistic Arithmetic
(Acta Universitatis Gothoburgensis, 2025-05-08) Granberg Olsson, Mattias
This dissertation concerns first-order theories of iterated positive truth and fixpoints over intuitionistic arithmetic in three respects: the strength of these theories relative to the arithmetic base theories, relationships between theories of positive fixpoints and compositional and disquotational truth for truth-positive sentences, and a comparison with the classical case. It is known that these theories over classical Peano arithmetic (PA) are mutually interpretable and exceeds the strength of PA. Over intuitionistic Heyting arithmetic (HA), on the other hand, finite iterations of strictly positive fixpoints have been shown to be conservative. After introducing the setting and presenting the earlier results, as well as the technical tools, the main section of the dissertation can be roughly divided into two parts. The first presents a novel proof of the conservativity result above. The proof interprets the theories into the logic of partial terms where a realizability interpretation is used to reduce the problem to fixpoints for almost negative operator forms. A diagonal argument using a hierarchy of almost negative formulae with corresponding partial satisfaction predicates yields the result. The second part generalises the tri-interpretation result from the second paragraph to intuitionistic theories, by proposing a new generalisation of positivity called guarded positivity with the aim to better capture the behaviour of intuitionistic implications and their interplay with transfinite iterations of truth and fixpoint predicates. As a corollary, these transfinite theories are stronger than HA. A discussion of the results and the methods used concludes the dissertation.
A Study of Actual and Non-Actual Motion Expressions in Telugu: A path towards a post-Talmian Motion Event Typology
(2024-11-05) Naidu, Viswanatha
This compilation thesis investigates the expressions of actual and non-actual motion (henceforth AM and NAM, respectively) in Telugu (Dravidian). AM refers to a situation where a physical object moves from one place to another, whereas NAM refers to a context where a motion verb describes a static scene. This semantics-based study is the first of its kind on Dravidian languages in general and Telugu in particular. In this respect, the main aim of this thesis is the detailed examination of the expression of AM and NAM in Telugu, leading to a proposal for a new Post-Talmain framework for studying motion event typology, thus advancing our understanding of semantic typology in general. Both AM and NAM data were collected from three separate but related experimental studies. Study 1 involved using the well-known wordless picture book Frog, where are you, to elicit the relevant data (AM data). Data for Study 2 was collected using video stimuli (AM data), and, for Study 3, data was collected using pictorial stimuli (NAM data). Study 1 made a comparative study of motion events in Telugu and Thai, demonstrating a number of serious challenges to and problems with Talmy’s binary typology. Therefore, we need to move beyond the Talmian terrain and enter what could be characterized as the post-Talmian era. In other words, this study sets the tone for a post-Talmian motion event typology, with languages such as Telugu and Thai data serving as case studies for the new proposal. In this new framework, languages are analyzed as belonging to clusters, with each cluster exhibiting specific properties and characteristics and it is further proposed that there are at least 4 possible clusters to which languages can belong. French and Swedish, for instance, represent one cluster each while Thai and Telugu represent the remaining two other clusters respectively, based on the fact that their linguistic encoding of motion is distinct from languages in the other clusters in many respects. Study 2 supports the proposal presented in Study 1 and empirically establishes the typological features of the fourth cluster to which Telugu belongs according to the post-Talmian motion typology proposed in Study 1. The characteristics of this cluster are frequent use of deictic verbs for MOTION, the predominant use of case markers for PATH, instead of path verbs, a dedicated set of spatial nouns for REGION, violation of the “boundary-crossing constraint”, and adnominal preponderance of coding over adverbal dominance. Study 3 examines NAM expressions in Telugu within the framework of the newly proposed post-Talmian typology. It establishes a noteworthy similarity between NAM and AM patterns, with the exception of the Manner of motion. This new framework identifies the use of generic deictic verbs for MOTION, case markers for PATH, infrequent use of path verbs, REGION expressing spatial nouns, and adnominal dominance as being common in both AM and NAM events in Telugu. The absence of manner of motion in NAM is in line with the literature and is in accordance with arguments that the encoding of NAM is to be explained by factors that are deeply rooted in human cognition. It is then proposed that these typological features may also be found in other South Asian and agglutinating languages which need to be investigated in future studies. Additionally, Study 3 explores the motivations underlying the use of NAM expressions, providing additional support for a previous claim that NAM is driven by multiple distinct experiences. In sum, the thesis contributes to the field of motion event typology by proposing a new post-Talmian motion typology, signaling the need for more languages to be explored along these lines to fully validate this relatively new proposal.
Computational Models of Language and Vision: Studies of Neural Models as Learners of Multi-Modal Knowledge
(2024-05-16) Ilinykh, Nikolai
This thesis develops and evaluates computational models that generate natural language descriptions of visual content. We build and examine models of language and vision to gain a deeper understanding of how they reflect the relationship between the two modalities. This understanding is crucial for performing computational tasks. The first part of the thesis introduces three studies that inspect the role of self-attention in three different self-attention blocks of the object relation transformer model. We examine attention heatmaps to understand how the model connects different words, objects, and relations within the tasks of image captioning and image paragraph generation. We connect our interpretation of what the model learns in self-attention weights with insights from theories about human cognition, visual perception, and spatial language. The three studies in the second part of the thesis investigate how representations of images and texts can be applied and learned in task-specific models for image paragraph generation, embodied question answering, and variation in human object naming.The last two studies in the third part examine properties of human-generated texts that multi-modal models are expected to acquire in image paragraph generation as well as perceptual category description and interpretation tasks. We analyse discourse structure in image paragraphs produced with different decoding methods. We also inspect whether models of perceptual categories can abstract from visual representations and use this knowledge to generate descriptions that exhibit discriminativity levels important for the task. We show how automatic measures for evaluating text generation behave in a comparison of model-generated and human-generated image descriptions. This thesis presents several contributions. We illustrate that, under specific modelling conditions, self-attention can capture information about the relationship between objects and words. Our results emphasise that the specifics of the task determine the manner and context in which different modalities are processed, as well as the degree to which each modality contributes to the task. We demonstrate that while favoured by automatic evaluation metrics in different tasks, machine-generated image descriptions lack the discourse complexity and discriminative power that are often important for generating better, human-like image descriptions.
Fundamental Dynamicity – A Metaphysics of Time and Process
(Acta Universitatis Gothoburgensis, 2024-05-02) Zachrau, Maximilian
This book continues a long tradition in philosophy going back to the pre-Socratics, who made a rather simple observation that still holds true today: All things flow. We experience this flow, flux, in abundance: the seasons coming and going, the sun rising and setting, plants, trees, children growing, ageing, and in many other respects. However, we also experience things persisting. We meet people again, visit places twice, wear clothes again, or live in the same city all our lives. That things flow, that they persist, and how they do both has been a (persistent) puzzle throughout the history of philosophy. The proposed answers come in abundance, but two central ideas stand out. One is to explain flux by what persists. Namely, flux is explained by change, which typically is understood to be grounded in substances and their properties. Accordingly, this explanation of flux is called substance metaphysics. On the other hand, there is process philosophy, which emphasises the pervasive dynamic character of reality. According to this view, dynamic entities like processes should be the basis for our conceptualisation of the world, especially flux; it claims that dynamic processes are fundamental. Process philosophy, which gained significant traction in the latter half of the 20th century, offers a compelling perspective that integrates scientific findings and everyday experience. However, it raises several crucial questions: What are these fundamental processes? How can we metaphysically understand them? And most importantly, what is dynamicity and how can we distinguish dynamic entities from static ones? I propose and defend answers to these questions, which connect to the philosophy of time, another influential philosophical topic. I advocate a view that posits dynamicity as a forward-directedness, suggesting that dynamic entities are inherently oriented towards the future. While this commits the process philosopher to accept the reality, directedness, and dynamicity of time, I argue vice versa that dynamic time views, in turn, require fundamental processes. What results is dynamic metaphysics—a package deal of process philosophy and dynamic views on the nature of time.
Look on my thesis, ye mighty: gaze interaction and social robotics
(2024-04-02) Somashekarappa, Vidya
Gaze, a significant non-verbal social signal, conveys attentional cues and provides insight into others’ intentions and future actions. The thesis examines the intricate aspects of gaze in human-human dyadic interaction, aiming to extract insights applicable to enhance multimodal human-agent dialogue. By annotating various types of gaze behavior alongside speech, the thesis explores the meaning of temporal patterns in gaze cues and their correlations. On the basis of leveraging a multimodal corpus of dyadic taste-testing interactions, the thesis further investigates the relationship between laughter, pragmatic functions, and accompanying gaze patterns. The findings reveal that laughter serves different pragmatic functions in association with distinct gaze patterns, underscoring the importance of laughter and gaze in multimodal meaning construction and coordination, relevant for designing human-like conversational agents. The thesis also proposes a novel approach to estimate gaze using a neural network architecture, considering dynamic patterns of real-world gaze behavior in natural interaction. The framework aims to facilitate responsive and intuitive interaction by enabling robots/avatars to communicate with humans using natural multimodal dialogue. This framework performs unified gaze detection, gaze-object prediction, and object-landmark heatmap generation. Evaluation on annotated datasets demonstrates superior performance compared to previous methods, with promising implications for implementing a contextualized gaze-tracking behavior in robotic interaction. Finally, the thesis investigates the impact of different gaze patterns from a robot on Human-Robot Interaction (HRI). The results suggest that manipulating robot gaze based on human-human interaction patterns positively influences user perceptions, enhancing anthropomorphism and engagement.
”A social engineer or a parasite on society”: The moral responsibility of enabling (un)ethical business conduct
(Acta Universitatis Gothoburgensis, 2024-02-28) Elliott, Jasmine Christine
Exploring Evidence-Based Practice Through New Forms of Engagement
(2023-12-15) Pistone, Isabella
This thesis is concerned with how to connect Science and Technology Studies (STS) with Evidence-Based Practice (EBP) through new forms of engagements. EBP is commonly associated with efforts to improve quality of welfare services. The principles and methods associated with EBP have been criticized for being reductionist. Such discussions pinpoint several challenges concerning principles for the production and utilization of evidence in EBP. At the same time, STS scholarship sheds light on informal practices that are often overseen in models and principles of EBP. In various ways, this research displays mismatches between epistemological assumptions underpinning EBP and empirical epistemologies at work when EBP is enacted by professionals in daily practice. In this thesis, I explore how such STS insights can be put to work for developing EBP. The thesis comprises five papers that work with different operationalizations of the guiding question: How can sensibilities from STS contribute to developments of EBP knowledge practices? The papers explore several domains of welfare where EBP principles have been adopted. Drawing on various sources of data such as interviews, observations, scholarly literature and situated experiments, these papers offer a diverse set of explorations into the current shapes of EBP and experimentation with how STS research can contribute with generative developments. Collectively, these papers challenge and expand the boundaries of EBP, offering a perspective that moves beyond narrow ideals of formalization and pre set knowledge hierarchies. Instead, they emphasize the dynamic interplay between various forms of knowledge necessary when EBP is to be realized in daily practice. Based on these papers, I outline characteristics for an epistemological reconceptualization of EBP that challenges the conventional usage of EBP as a descriptor for standardized interventions. I discuss how the experimentation with STS approaches renegotiates roles, positions, and engagements of STS researchers. I conclude by showing how the engagement in this thesis contribute to an expansion of boundaries; not only boundaries around EBP, but boundaries around STS scholarship as well.
Relaxing Normativity: Essays on Relaxed Approaches to Realism about Normativity
(Acta Universitatis Gothoburgensis, 2023-12-14) Karimi, Paiman
Nonhuman Moral Agency: A Practice-Focused Exploration of Moral Agency in Nonhuman Animals and Artificial Intelligence
(2023-12-05) Behdadi, Dorna
Can nonhuman animals and artificial intelligence (AI) entities be attributed moral agency? The general assumption in the philosophical literature is that moral agency applies exclusively to humans since they alone possess free will or capacities required for deliberate reflection. Consequently, only humans have been taken to be eligible for ascriptions of moral responsibility in terms of, for instance, blame or praise, moral criticism, or attributions of vice and virtue. Animals and machines may cause harm, but they cannot be appropriately ascribed moral responsibility for their behavior. This thesis challenges the conventional paradigm by proposing an alternative approach where moral agency is conceived as the competence to participate in moral responsibility practices. By shifting focus from intra-individual to contextual and socially situated features, this practice-focused approach appears to make the attribution of moral agency to nonhuman animals and AI entities more plausible than commonly assumed. Moreover, considering the current and potential future prevalence of nonhuman animals and AI entities in everyday settings and social contexts, a potential extension of moral agency to such entities could very well transform our social, moral, and legal practices. Hence, this thesis proposes that the attribution or withholding of moral agency to different entities should be carefully evaluated, considering the potential normative implications.
Studies in Language Structure using Deep Learning
(2023-08-15) Ek, Adam
This thesis deals with the discovery, prediction, and utilization of structural patterns in language using deep learning techniques. The thesis is divided into two sections. The first section gives an introduction to the tools used and the structures in language we are interested in. The second part presents five papers addressing the research questions. The first three papers deals with discovering and predicting patterns. In the first paper, we explore methods of composing word embeddings to predict morphological features. The second paper deals with predicting the depths of nested structures. The remaining three papers deal with using structures in language to make semantic predictions. The third paper explores using dependency trees to predict semantic predicate-argument structures using a rule-based system. The fourth paper explores modeling linguistic acceptability using syntactic and semantic labels. The fifth paper deals with exploring how punctuation affects natural language inference.
Semantic change in interaction: Studies on the dynamics of lexical meaning
(2023-03-28) Noble, William; Noble, Bill
This compilation thesis investigates how word meanings change. In particular, it's concerned semantic change at the levels of interaction and the speech community. To this end, the compiled studies employ methods from both formal and computational semantics. The first study presents a model for, and companion annotation study of, word meaning negotiation, a conversational routine in which the meaning of a word becomes an explicit topic of conversation. The next two studies introduce and apply classification systems, a model of communal conceptual resources for ordering and talking about a particular domain. We use a formalization thereof to model how genus-differentia definitions can be used in interaction to update lexical knowledge of perceptual categories. The next study considers a related phenomenon, perceptual category description, but this time from a computational perspective. By modeling a short interaction between two neural networks, we investigate how different ways of representing perceptual categories affect linguistic grounding. Following that, we turn to the dynamics of social meaning, particularly the meaning of implicit conversational assumptions called topoi, with a focus on situations of involving uncertainty about the speaker's social identity. The final two studies of the thesis shift the focus from particular interactions to the level of the community. First, we investigate linguistic variation using community conditioned language models to learn vector representations for a collection of online communities. These language-based representations are found to correlate with community representations based on community membership alone. Finally, we use diachronic distributional word vectors to study short-term semantic shift in online communities. We find that semantic change has a significant yet nuanced relationship with the social structure of the community. Altogether, the compilation offers two main insights. First, semantic plasticity is directly related to the complexity of the lexical semantic system. Words exhibit both perceptual and inferential meaning potential, each of which play a role in conveying and learning new meanings. Monolithic representations of word meaning belie a structured flexibility that guides how words can be used, while providing opportunities for innovation. It is this flexibility that is often the site of new conventionalized meanings. Second, semantic change is rooted in the interactive practices of the community. Communities sustain the communicative norms that govern how linguistic interaction takes place. These norms also provide a framework for negotiating meaning, and comprise the social and semiotic context that supports semantic innovation and change.
Peerage and Judgment: How transdisciplinary collaborations recognize contributions without a consensus of meaning
(2022-12-07) Lundgren, Jakob
This thesis concerns judgments of quality and belonging in transdisciplinary research (TD). TD includes academics from various disciplines and is open to participation from non-academics. It typically aims to address societal problems and is argued to produce knowledge that is more nuanced than traditional disciplinary research due to the plurality of perspectives included. The focus of this thesis is on the dynamics underlying judgments made by TD collaborations where members recognize each other as epistemic peers despite different conceptions of what it means for science to be good. To investigate these dynamics, I adopt a middle perspective that connects theoretical and empirical investigations. This middle-level theory illuminates two issues surrounding epistemic peerage in TD. The first issue concerns the coordination of the demarcation of a TD collaboration and the collaboration across boundaries within the collaboration. Boundaries are drawn towards an outside of non-peers while the peers within the collaboration maintain a multiplicity of understandings. Central is that those within the collaboration cannot have world-views that are so different as to prevent them from recognizing each other as peers, while also not so similar that there can be no substantial exchanges across borders. I show how the investigated cases use hub-and-spoke concepts to coordinate demarcation and collaboration. The second issue concerns which issues are kept open and closed for discussion within a TD collaboration. The aims of TD of production of nuanced knowledge with societal relevance and inclusive practices require an openness to discuss matters that would in other circumstances be considered closed facts. At the same time a certain amount of closedness is required to stabilize the collaboration. The cases in this thesis show how the question of which issues are kept open and closed is affected by the institutional environment of TD collaborations.
Weapons of Mass Destruction: Financial Crises from a Philosophical Perspective
(2022-11-09) Endörfer, Richard
Financial crises are severely destructive events. The Global Financial Crisis of 2008 sent sovereign states into a spiral of political unrest and caused millions of people to lose their homes, their jobs, their life savings, their health, and in many cases even their lives. But financial crises are not unavoidable natural events. They are the consequences of intentional human behaviour. To be more precise, they are unfortunate side-effects of everyday financial practices. If these practices are not carefully monitored and reined in, they can, in words borrowed from Warren Buffet, become “weapons of mass destruction”. This thesis is an attempt at an interdisciplinary investigation of financial crises. It combines arguments from normative ethics, political philosophy, economics and law in order to discuss three questions at the heart of the public debate on financial crises: “Who is responsible for bringing about financial crises?”; “What precisely is wrong with practices that contribute to the risk of financial crises?”, and “What can be done to mitigate the risk of financial crises?”. A few key insights offered in this thesis are as follows: First, financial crises do not emerge because of the misbehaviour and greed of a few “bad apples”, rather, they are the result of “business as usual” within financial markets. Second, there are strong reasons for states to regulate financial markets heavily in order to prevent severe harm. Third, there are few good reasons to believe that consumers can be held morally responsible for contributing to financial crises.
Who is laughing now? Laughter-infused dialogue systems
(2022-08-16) Maraev, Vladislav
This thesis paves the way for including laughter in spoken dialogue systems in a domain-general and linguistically valid way using computational linguistics tools and methods. The thesis is concerned with three main areas. The first area concerns the placement of laughter in relation to speech and other behaviours. We show that convolutional and recurrent neural networks can effectively predict laughs from dialogue transcripts, whereas human perforance in this task is significantly worse. Such models allow dialogue systems to predict user laughter and, if needed, put system laughter in an appropriate place. Further, we look at laughter placement in relation to gaze and show that laughter performing different pragmatic functions is related to different gaze patterns. These findings provide important implications for embodied conversational agents and social robots in regard to multimodal behaviour realisation and coordination. The second area is concerned with the interaction between laughter and the communicative intent of a user and system, as well as with the context in which the given intent occurs. We lay the groundwork for the central component of a spoken dialogue system by implementing a dialogue manager in a theoretically informed way using a proof-theoretic model based on linear logic. Our dialogue manager is then extended to support laughter functioning as feedback or as a signal accompanying system feedback, and an answer to polar questions. Additionally, we look at how laughter can modify or shape a dialogue act, and how the inclusion of laughter can improve Transformer-based deep learning models in the task of dialogue act recognition. The third area is humour. Even though humour is not necessary for laughter, they are closely related. We look at how humour is related to reasoning about social conventions and other learned and accommodated implicit assumptions. We show how these can be modelled informally, suggesting one way in which the goal of situational and conversational creativity can be realised in artificial agents.
Confluence and Divergence of Emancipatory Healthcare Ideals and Psychiatric Contextual Challenges
(Acta Universitatis Gothoburgensis, 2022-05-09) El-Alti, Leila
Person-centered care (PCC) is generally understood to involve shaping healthcare processes, decisions, and plans according to the individual values, preferences, or goals of each patient. This is in contrast to more traditional approaches which provide care to patients based on standard clinical guidelines. In healthcare and bioethical literature, PCC is often praised as an ideal approach of healthcare provision because it is thought to empower patients and improve their adherence, satisfaction, and overall health outcomes. However, the notion has been defined in different ways, and it is unclear how and whether it can be implemented in all healthcare settings. This dissertation aims to elucidate the concept of PCC and explore the implications of its intersection with psychiatry. The work contextualizes the concept within larger healthcare and social movements, and in that light, analyzes its values, decision-making process, and ambitions. The unique and complex challenges that psychiatric care settings engender are further used to examine how PCC commitments fare when faced with the limitations of mental illness and restrictive conditions of psychiatric facilities.
Resources and Applications for Dialectal Arabic: the Case of Levantine
(2022-05-03) Qwaider, Chatrine; Abu Kwaik, Kathrein
This is a thesis about the computational study of Dialectal Arabic (DA). In particular, the thesis studies DA, with a special emphasis on Levantine Arabic, and develops tools and resources for the computational study of Dialectal Arabic Natural Language Processing (DANLP). It investigates the creation of fine-grained resources that can be used for a variety of computational tasks, and a number of effective models that can deal with the complexity of fine-grained dialectal data. Dialect Identification (DI), as well as Sentiment Analysis (SA) are the Natural Language Processing (NLP) tasks investigated in this thesis. In the first part (Study 1 and Study 2), I study the DI task on both coarse-grained and fine-grained levels. For this reason, I build the first annotated Levantine (SHAMI) Dialect Corpus (SDC). Furthermore, I explore the ability of n-gram language models, Machine Learning (ML) algorithms and ensemble learning techniques to classify and detect 26 Arabic varieties. In the second part, I conduct a linguistic study to measure the lexical distance between MSA and DA, and between the dialects themselves. This is done to check whether transferring knowledge from one variety to another is possible. In the third part, studies 4,5 and 6, I explore Arabic Sentiment Analysis (SA). I investigate the idea of knowledge transfer between MSA and the dialects using SA as a case study. Furthermore, I implement various models such as the pre-trained language model BERT, Deep Learning (DL), ML and feature engineering approaches to detect the sentimental polarity of DA data. I introduce two valuable resources for this task, one focusing on Levantine sentiment (Shami-Senti), and the other for DA in general (ATSAD). I exploit different ways of annotation, e.g. human, lexicon-based and automatic distant supervision annotation. The last study is about choosing the best model for DI and SA. I exploit well-known models and approaches using various kinds of DA resources. The thesis contributes to the field of DANLP in a number of ways. The introduced valuable resources can be seen as a stepping stone for a deeper investigation and understanding of issues in DANLP. They are also reliable and can be used by researchers to address different NLP tasks. The cross-dialectal linguistic studies will open up prospects for researchers to fine-tune models and transfer knowledge among Arabic varieties. A big part of the contribution lies in designing DI and SA models. I implement several ML models that use feature engineering approaches and N-gram language models to identify the dialect or detect the sentiment. For DI, I design and implement an ensemble learning model that is able to handle fine-grained dialects. Additionally, I exploit the usage of DL models on different SA dialectal datasets and achieve competitive results. For both tasks, I exploit the recent pre-trained language models and perform a comparison to choose the best model. I also implement a semi-supervised approach for automatic labelling and annotating data with the help of self-training techniques to improve the performance of the dataset. These models will help researchers dive deeper into DANLP and create practical and industrial systems.
Giving Executives Their Due: Just Pay, Desert, and Equality
(Acta Universitatis Gothoburgensis, 2021-11-19) Andersson, Alexander
Before, during, and after the global financial crisis of 2008, executive pay practices were widely debated and criticized. Economists, philosophers, as well as the man on the street all seem to have strong feelings towards how much, in what ways, and on what grounds executives are paid. This thesis asks whether it is possible to morally justify current executive pay practices and, if so, on what grounds they are justified. It questions those who find no quarrel with pay practices due to their minimally moralized view of the market and – perhaps more importantly – it asks for a sophisticated critique from those who criticize current pay practices. The discussion on just pay is clearly one of distributive justice. One reason for why some people consider CEO pay practices to be (fairly) unproblematic while others find them objectionable stems from the availability of different understandings of, and principles for, just pay. We tend to associate justice in pay with such things as proper incentivization, being the outcome of a fair procedure, being deserved on the basis of effort or contribution, and/or satisfying the ideal of equality. Parts of this thesis are devoted to these different understandings and what they entail for the moral justification of CEO pay. Another reason for the conflicting views on CEO pay stems from how issues of justice go beyond the confines of economics and applied ethics, extending all the way into the domain of political philosophy. Parts of this thesis explore this connection, in particular how the concept of economic desert relates to the broader concept of moral desert. Lastly, I discuss the criticism that the superrich (including executives) are being paid too much and are in possession of too much wealth. The issue at hand here is how to morally justify the interventions that seem suitable to rectify the situation.
Explorations of the Relationship Between the right to Make Decisions and Moral Responsibility in Healthcare
(2020-11-18) Hartvigsson, Thomas
People intuitively think that there is a strong connection between having a right to make decisions and to be morally responsible for those decisions. This thesis explores the relationship between these notions in the context of healthcare. The exploration particularly focuses on what I call fringe decisional agents, e.g. adolescents and people who suffer from mental disorder, who have uncertain decision-making competence and exist at the intersection of different institutions. I argue that even though the two notions are strongly connected they can come apart. First, even though both notions are concerned with the moral status of a person there is a potential conflict between the appropriate responses to a person who has the right to decide and someone who is morally responsible. Second, even if conditions for having the right to decide and being morally responsible are very similar they can come apart. Moral responsibility requires that a person exercises a certain degree of control over their actions, a condition that has no clear equivalent for the right to decide. Furthermore, even though both have cognitive conditions, the condition for having the right to decide is directed towards information regarding oneself, whereas the condition for moral responsibility is primarily directed towards information about other people. Finally, if an agent is the concern of different institutions, these might have different conditions for assigning the relevant status and may furthermore do so at different times.
Natural Language Processing for Low-resourced Code-switched Colloquial Languages – The Case of Algerian Language
(2020-06-09) Adouane, Wafia
In this thesis we explore to what extent deep neural networks (DNNs), trained end-to-end, can be used to perform natural language processing tasks for code-switched colloquial languages lacking both large automated data and processing tools, for instance tokenisers, morpho-syntactic and semantic parsers, etc. We opt for an end-to-end learning approach because this kind of data is hard to control due to its high orthographic and linguistic variability. This variability makes it unrealistic to either find a dataset that exhaustively covers all the possible cases that could be used to devise processing tools or to build equivalent rule-based tools from the bottom up. Moreover, all our models are language-independent and do not require access to additional resources, hence we hope that they will be used with other languages or language varieties with similar settings. We deal with the case of user-generated textual data written in Algerian language as naturally produced in social media. We experiment with five natural language processing tasks, namely Code-switch Detection, Semantic Textual Similarity, Spelling Normalisation and Correction, Sentiment Analysis, and Named Entity Recognition. For each task, we created a dataset from user-generated data reflecting the real use of the language. Our experimental results in various setups indicate that end-to-end DNNs combined with character-level representation of the data are promising. Further experiments with advanced models, such as Transformer-based models, could lead to even better results. Completely solving the challenge of code-switched colloquial languages is beyond the scope of this experimental work. Even so, we believe that this work will extend the utility of DNNs trained end-to-end to low-resource settings. Furthermore, the results of our experiments can be used as a baseline for future research.

Browse

Recent Submissions