CNI College

This dissertation investigates variation in linguistic framing of events. People differ in how they describe the same event, and this raises the question of how such variation can be systematically measured in textual data and how it can be explained. The study builds on the fundamental assumption that meaning arises from two complementary processes: first, the referential grounding of an expression, which identifies which event in the world the expression refers to; and second, the conceptual disambiguation of the expression, which determines which frame it evokes. While both processes have long been recognized in linguistic research, they have not previously been modeled together on a large scale in textual data. This dissertation demonstrates how such an integrated approach can be constructed and used for empirical analysis. To achieve this, the research combines two theoretical resources. For referential grounding, it uses structured data from digital ontologies such as Wikidata, which contain information about events, their participants, locations, times, and types. For framing, it draws on FrameNet, a lexicographic database of more than a thousand semantic frames linked to lexical units and enriched with specifications of semantic roles. Using the Multilingual Wiki Extraction Platform, the study compiles a corpus in which event types are represented through large sets of incidents extracted from Wikidata, accompanied by texts sourced from the bibliographies of relevant Wikipedia pages. These texts are annotated in the Dutch FrameNet annotation tool with a two-layer system: entity linking connects expressions to the structured data of the events they denote, while frame annotation identifies which frames the same expressions evoke. The result is a corpus in which large numbers of texts are grouped under shared events, enriched with information on both their referential links and their framing structures. This corpus makes it possible to measure framing variation at a fine-grained level. The question then becomes how to explain this variation. To address this, the dissertation develops a pragmatic framework centered on the role of common ground between writers and readers. The key idea is that the amount of shared knowledge about an event influences how it is described. When common ground is low, as immediately after an event, writers tend to provide extensive and varied descriptions, resulting in high framing variation. When common ground is high, as after time has passed, minimal descriptions are sufficient, resulting in low variation. Over time, therefore, variation decreases. Yet when new related events occur, they generate fresh narratives. Writers focus on the new developments with many and varied frames, while the initial event, now familiar, is referred to only briefly and in less diverse terms. From this framework, the dissertation formulates hypotheses about framing variation as a function of common ground over time. These hypotheses concern frames, lexical units, constructions, and interfaces, and they are tested in the referentially grounded FrameNet corpus at the levels of event type, incident, and participant. The contribution of this work is both empirical and theoretical. Empirically, it develops a large-scale corpus that integrates structured event data with FrameNet-based annotation, providing a new resource for corpus and computational linguistics. Theoretically, it offers a pragmatic model that links framing variation to dynamics of common ground and storytelling, shedding new light on how events are represented and narrated over time.

In a world where conversational Artificial Intelligence (AI) is playing an increasingly large role in our daily lives, it is important to understand how conversations between humans and AI or robots can become more personal, and stay useful over a prolonged period of time. The way two humans talk about other people is shaped by their degree of shared knowledge, or common ground. These referring expressions tend to become shorter and more personal over time, as codified, shared ways of making reference called conventions start to form. This process could make conversations with AI such as robots more personal, but there has been little research on this topic. In this dissertation, I examine how common ground plays in a role in making reference in this Human-Robot Interaction (HRI), addressing the research gap that exists in this space, and providing insights into both human-human and human-robot use of referring expressions in longitudinal interactions. My investigation has three main goals: examining how useful data can be obtained to study referring expressions in HRI; examining how referring expressions in HRI are influenced by the common ground, conventions, and contextual factors such as ambiguity; and examining the role of the robot in shaping, using and understanding referring expressions and conventions. I carry out this investigation through a number of experiments using data sets, field experiments with a physical robot, and online experiments with a chatbot. My approach draws from fields in linguistics related to (human-human) common ground and pragmatics, as well as fields in Natural Language Processing related to Natural Language Understanding and Referring Expression Generation. An important element in my research is the introduction of an inner circle of familiar entities, for which common ground is built up, and an outer circle of unfamiliar entities. By comparing the use of referring expressions for the inner and outer circle, I create a contrastive analysis which allows me to draw conclusions about the impact of common ground on referring expressions. This contrast is implemented in an experimental framework called SPOTTER, a game in which referring expressions to the inner and outer circle are elicited. I use this framework to analyse the use of referring expressions and the role of the robot in shaping them. One of the main findings of this dissertation is that, although present, the role of common ground is less pronounced in human-robot reference-making compared to human-human interactions. However, I do find a large influence of the contextual setting within each round of the SPOTTER game, as shaped by the outer circle. This influence is shown by longer, more detailed referring expressions when ambiguity is high, as well as through longer referring expressions for the outer circle compared to the inner circle. The findings also suggest that the robot's use of referring expressions has a large impact on which referring expressions humans use. However, though both humans and robots form conventions, they rarely truly converge on a shared convention. Naturally, the findings are influenced by design choices for the experimental framework and the conversational model, as well as by the complicated nature of field studies in HRI. The dissertation lays the groundwork for further investigations of reference-making and common ground in HRI, providing experimental tools and models.

In recent years, numerous research studies have highlighted that CPU microarchitecture designs are crucial not only for performance but also for security. The release of Spectre and Meltdown in 2018 marked a significant turning point, revealing how microarchitectural optimizations could unintentionally introduce critical vulnerabilities that are exceptionally difficult to address due to the immutability of hardware after production. This thesis conducts an in-depth exploration of CPU microarchitectures, uncovering new attack surfaces and proposing optimizations to enhance defenses. Specifically, it evaluates the limitations of existing hardware mitigations against Spectre, demonstrating their insufficiencies and exposing pathways for continued exploitation. Furthermore, the research identifies previously unexamined root causes of transient execution vulnerabilities, leading to the discovery of new classes of attacks. In addition, by addressing key microarchitectural bottlenecks, this work introduces a novel way to accelerate memory error detection, a crucial aspect of software security assessment. By bridging the gap between hardware and software perspectives, this research deepens our understanding of CPU microarchitectures, paving the way for more secure and efficient computing systems.

Computers are complicated, and the security of modern software often depends on programmers anticipating and mitigating a dizzying assortment of vulnerability classes. Both researchers and adversaries continue to add additional complications based on hardware behavior to this mix, such as data-dependant power consumption, fault injection, cache timing attacks and transient execution - adding new attack surface which needs to be taken into account when building secure systems. In this dissertation, we look beyond the traditional attack surface, and find that many mitigations - across a variety of layers, from hardware to software - are compromised by faulty assumptions about the actual behavior of hardware. We begin close to the hardware, discovering that the relationship between two different forms of hardware attack - Fault Injection and Side-Channel Analysis - is so strong that we can find ways to apply power side-channel attack techniques directly to the results of voltage fault injection campaigns, without the need for an attacker to obtain power measurements. Next, we move to microarchitectural attacks - which rely on hardware behavior, but can be performed by software. By combining transient execution with unexpected - and unintended - CPU pipeline behavior, we show that the scope of software side-channel attacks can be expanded far beyond previous work, allowing both in-flight and 'stale' data to be stolen from the pipeline by an attacker, bypassing software mitigations by ignoring security domains entirely. Such attacks can be mitigated by isolating workloads on different cores - which do not share a pipeline. Unfortunately, we find that even this level of isolation can fail to prevent attackers from being able to leak critical data using side-channel attacks - such as random numbers used when generating cryptographic keys. Hence, these attacks must be mitigated in hardware. Finally, we investigate defenses against transient execution attacks such as Spectre which depend only on well-understood hardware behavior, but require software mitigations. We show that we can build efficient and practical software-based defenses which not only mitigate many Spectre attacks, but even harden software against architectural information leaks.

Abstract Standard Description Logics (DLs) can encode quantitative aspects of an application domain through either number restrictions, which constrain the number of individuals that are in a certain relationship with an individual, or concrete domains, which can be used to assign concrete values to individuals using so-called features. These two mechanisms have been extended towards very expressive DLs, for which reasoning nevertheless remains decidable. Number restrictions have been generalized to more powerful comparisons of sets of role successors in $$\mathcal {ALCSCC}$$ ALCSCC , while the comparison of feature values of different individuals in $$\mathcal {ALC} (\mathfrak {D})$$ ALC ( D ) has been studied in the context of $$\omega $$ ω -admissible concrete domains $$\mathfrak {D}$$ D . In this paper, we combine both formalisms and investigate the complexity of reasoning in the thus obtained DL $$\mathcal {ALCOSCC}(\mathfrak {D})$$ ALCOSCC ( D ) , which additionally includes the ability to refer to specific individuals by name. We show that, in spite of its high expressivity, the consistency problem for this DL is ExpTime-complete, assuming that the constraint satisfaction problem of $$\mathfrak {D}$$ D is also decidable in exponential time. It is thus not higher than the complexity of the basic DL $$\mathcal {ALC}$$ ALC . At the same time, we show that many natural extensions to this DL, including a tighter integration of the concrete domain and number restrictions, lead to undecidability.

近年，生成AIや大規模言語モデルの急速な発展により，私たちの生活や仕事の在り方が大きく変容しつつあります。こうした技術革新の波の中で，「人」と「AI」の関係性をより良く構築するためには，人間の行動や心理に関する深い理解が欠かせません。

We are increasingly confronted with complex societal issues such as climate change or social segregation. Higher education institutes (HEIs) have a crucial role in addressing such issues as part of their so-called third mission activities. In addition to the more market-driven forms of (technological) knowledge transfer, these complex issues require collaborative processes in defining and addressing their dynamic nature. The thesis focuses on engaged education within HEIs, as way to giving meaning to these collaborative processes through community engagement. To gain more insight into the implementation and sustainable embedment of such practices we aim to answer the following research question; How to embed engaged education within HEI, in order to contribute effectively to addressing (complex) societal issues? Our approach takes a multi-level perspective, building on the framework of Bogel et al. (2019), who combine transition studies and organisational change management. Through this framework, we allow for an explicit analysis of the role of individuals at the micro level but also consider organisational, meso-level processes when embedding engaged education. We conducted a reflection stream together with practitioners in order to gain insights into their rationales for embedding engaged education. Moreover, a theoretical stream broadly explored embedding engaged education at the individual course (micro) level as well as at the organisational (meso) level. Finally, two case studies, using an action–research approach, were conducted to gain insights into cross-disciplinary approaches for addressing complex societal issues through engaged education. Through these studies we explored systematic strategies and cross-disciplinary approaches for embedding engaged education to effectively address (complex) societal issues. Our findings show the potential of engaged education to centralise community values, needs and interests and thereby increase HEIs’ societal impact. We outline some of the tensions that can arise between individual action and organisational practices when embedding engaged education that centralise community engagement. To deal with these tensions, we describe systematic strategies for embedding engaged education for addressing societal issues at the individual course or micro level, as well as at the organisational meso level. From this we distil two guiding principles that we deem crucial when centralising community engagement as part of engaged education, namely catering for 1) continuous faculty, community and student involvement and 2) flexibility and continuity of engaged education activities. In relation to these guiding principles, we emphasise the importance of individual roles of faculty, community stakeholders and students at the micro level, but also recognise that the process of embedding engaged education, by definition, cannot rely solely on them. Changes in the organisational culture and in its policies and practices at the meso level are essential. Alignment between micro- and meso-level practices, through these guiding principles, can contribute to the institutional coherence needed to centralise community values, needs and interests. Building on this notion, we propose two concrete cross-disciplinary approaches to facilitate engaged education for addressing complex societal issues. Given their faculty-spanning nature these approaches position themselves as ‘transition initiatives’ at the meso level, while also connecting to the micro level, thereby facilitating the process of institutional coherence. We outline how such transition initiatives might contribute to transformation within and beyond individual HEIs. We believe that this thesis contributes to placing more emphasis on societal engagement, advocating for more intense involvement of community stakeholders within education. In this way, we provide valuable insights into how HEIs can purposely give meaning to their third mission activities and make societal impact through engaged education.

Metaphor scholars have widely explored metaphor use in political discourse. Nevertheless, the current research does not account for the ‘gradable metaphoricity’ in political discourse analysis. This dissertation fills this gap by addressing this specific issue in two frameworks: (1) viewing political metaphor from a static and gradient perspective (Source-Target mapping; Conventional vs. Novel vs. Dead), and (2) viewing political metaphor from a gradable and dynamic perspective (a matter of salience and awareness of metaphoricity). A systematic literature review in chapter 2 points out that the static and dynamic perspectives differ significantly in underlying assumptions and organizing principles, although both are indistinctly referred to by metaphor scholars as constituting a ‘gradable’ view. The former takes metaphor as a static conceptual unit or lexical unit, but the latter tends to accord a central role of activation of metaphoricity to metaphorical expressions. To launch a theoretical advancement about the dynamic view in political discourse, chapter 3 offers a usage-based model of gradable and dynamic metaphors—the YinYang Dynamics of Metaphoricity (YYDM). In addition, this thesis investigates political metaphors from an interdisciplinary angle, incorporating theory from the field of International Relations. An empirical evaluation of political (discourse) studies in chapter 4 shows the large absence of transdisciplinary perspectives. Addressing the abovementioned gaps, this dissertation reports on two empirical analyses of trade metaphors in a big corpus that represents the official trade positions of the United States and China during the presidencies of Bill Clinton and Jiang Zemin (1993-1997) as well as Donald Trump and Xi Jinping (2017-2021). Based on a codebook of a cross-linguistic metaphor identification procedure in chapter 5, the first empirical part contributes to the static and gradient perspective and includes two corpus-based studies of metaphorical framing about trade (chapters 6-7). The diachronic and cross-linguistic use of source domains from a socio-cognitive approach in chapter 6 reveals that source domains are semantic fields that vary with trade discourse contexts (interests, power, and power relations). Chapter 7 shows that the use of trade metaphors (source domains of Conventional and Novel metaphors) to construct and legitimize political ideologies correlates with differences between political genres. The second part contributes to the gradable and dynamic view by applying the transdisciplinary model of YinYang Dynamics of Metaphoricity in chapters 8-10. In chapter 8, an evaluation of the new model in the Clinton-Jiang trade discourse shows that the dynamic cognitive process (transformation of metaphoricity) and rhetorical process (argumentation and persuasion) mutually develop with the evolution of the socio-political process (trade perspectives and trade events). Chapter 9 investigates the transformation of metaphoricity in the Trump-Xi trade discourse and finds that cognitive processes (patterns of metaphoricity activation) and affective processes (emotions or sentiments) mutually develop with the evolution of socio-political processes (trade perspectives and trade events). Based on the findings in chapters 8-9, chapter 10 further shows several phenomena in the Clinton-Jiang and Trump-Xi trade discourses: the movement of metaphors on the metaphoricity spectrum, the bodily motivation of gradable and dynamic metaphoricity, and the interconnected political discourse systems. Drawing on all the theoretical and empirical insights revealed in the dissertation, the final section of the thesis outlines a future direction, i.e., moving towards a transdisciplinary and dynamic approach to metaphor in political discourse analysis.

is onderdeel van het proefschrift PDF zoals upgeload (thesis Klazien Tilstra zie boven) hoofdstuk 8, 8.1