UWSpace

UWSpace is the University of Waterloo’s institutional repository for the free, secure, and long-term home of research produced by faculty, students, and staff.

Depositing Theses/Dissertations or Research to UWSpace

Are you a Graduate Student depositing your thesis to UWSpace? See our Thesis Deposit Help and UWSpace Thesis FAQ pages to learn more.

Are you a Faculty or Staff member depositing research to UWSpace? See our Waterloo Research Deposit Help and Self-Archiving pages to learn more.

Photo by Waterloo staff

Communities in UWSpace

Select a community to browse its collections.

Recent Submissions

  • Thumbnail Image
    Item type: Item ,
    Development and Evaluation of Assistive AI Systems for Assessing News Trustworthiness
    (University of Waterloo, 2026-08-04) Zhang, Dake
    Online news shapes how people form opinions on topics such as science, health, and politics, yet the same environment that makes it widely accessible also enables low-quality and deceptive content to propagate at scale. This thesis studies how AI systems can help readers assess the trustworthiness of online news. Rather than predicting whether an article is true or false, we treat trustworthiness assessment as a process to be supported: a useful system should help readers ask the investigative questions that a careful reader would ask, and synthesize the external evidence and context needed to answer them. This framing is grounded in lateral reading, the strategy professional fact-checkers use when evaluating online information by searching beyond the page itself. We pursue system development and evaluation together because the two are inseparable in a new research area. This research program began with ReadProbe, a proof-of-concept retrieval-augmented LLM system that generated investigative questions, retrieved web evidence, and produced attributed answers. A subsequent pilot study, originally intended as a formative step toward a larger human-baseline study of question generation, characterized the kinds of questions university-affiliated readers wrote before and after brief lateral-reading guidance, and revealed useful design lessons. These lessons motivated the TREC 2024 Lateral Reading Track, the first shared benchmark in this area, which formalized question generation and document retrieval as foundational tasks. Results showed that question generation remains challenging even for frontier LLMs, while document retrieval is comparatively mature. A follow-up analysis further revealed that LLM-generated question lists are less diverse than those written by human experts and overlap little with them, suggesting that alignment with expert investigative priorities is the core open problem. These observations led to the TREC 2025 DRAGUN Track, which shifted the benchmark toward a more reader-oriented setting by introducing report generation as the main task and replacing direct grading with expert-authored, importance-weighted rubrics built through open-web research. To support the track, we developed an iterative multi-agent RAG system that simulates a lateral reader by interleaving query generation, multi-stage segment retrieval, information sufficiency evaluation, question generation, and report writing, which served as a strong starter-kit baseline. Finally, to make rubric-based evaluation reusable beyond the originally judged runs, we released an LLM-based AutoJudge system that mirrors the human assessment protocol and preserves run-level rankings well against the official human judgments. Through this research journey, three key findings stand out. First, generating investigative questions that align with expert priorities remains hard: even the strongest system in the DRAGUN track covered only about one-third of the importance-weighted rubric question space on average. Second, given a useful question, document retrieval is no longer the central bottleneck. The harder problems now sit at the planning and synthesis ends of the pipeline. Third, evaluation matters as much as system design, and rubric-based evaluation grounded in expert open-web research is a more useful and interpretable target than direct grading of system outputs. Overall, this thesis contributes a reader-centered framework for assistive AI in news trustworthiness assessment, along with shared tasks, datasets, systems, and a reusable evaluator that together support continued progress toward AI tools that help readers think more carefully about online news rather than think for them.
  • Thumbnail Image
    Item type: Item ,
    Evolving Trade-offs Towards Deployable Private Systems for Data Science
    (University of Waterloo, 2026-08-04) Humphries, Thomas
    There is no one-size-fits-all solution to preserving privacy in data science. While insights derived from sensitive data can benefit society, privacy is typically at odds with utility, performance, usability, or some combination of these objectives. Furthermore, each system differs in its definition of these objectives and the way they interact with one another. If the compromise required for any single objective is too great, the system will not be deployed, or worse, will be deployed with a weakened privacy guarantee, exposing users to potential harm. In this work, we address this challenge from multiple angles. First, through strategic algorithm design, our work creates private systems with improved trade-offs, enabling their deployment. This includes a more efficient protocol for the secure inference of deep machine learning models, a novel construction for aggregating key-value data in the local trust model, an evolutionary approach to improve the utility of private clustering, and a user-friendly interpretation of the error of private median queries. Second, we audit private systems to show the privacy risks associated with misleading privacy claims. In particular, through a privacy audit of machine learning, we highlight a difference between expectation and reality in privacy protections.
  • Thumbnail Image
    Item type: Item ,
    From Vulnerability to Viability in Small-Scale Fisheries: A Thematic Synthesis on Justice, Knowledge, and the Making of Transition
    (University of Waterloo, 2026-08-04) Shetu, Shahriyer Hossain
    Small-scale fisheries (SSF) sustain the livelihoods, food security, and cultural identities of approximately 500 million people globally, yet they face profound and compounding vulnerabilities arising from climate change, colonial dispossession, institutional exclusion, and the displacement pressures of an expanding blue economy. Despite their global significance, no systematic synthesis of the knowledge produced through the V2V Global Partnership Thematic Webinar Series has previously been conducted, leaving five years of transdisciplinary expert discourse inaccessible as a coherent scholarly contribution. This qualitative study addresses that gap through a reflexive thematic analysis of all 59 V2V thematic webinar sessions spanning January 2021 to December 2025, conducted within a constructivist paradigm and guided by an original tri-layered conceptual framework called the V2V Transition Ecosystem, which integrates the V2V vulnerability-to-viability core axis, the blue justice evaluative lens, and social-ecological systems theory. The analysis identified 74 codes across three research questions, organized into Core, Emerging, and New classifications and tracked temporally across five years. Five key findings emerge: the V2V webinar series documents a paradigmatic shift from vulnerability management to justice assertion; the five vulnerability types present across all five years are structural features of the global political economy of fisheries rather than developmental deficits; the eight Core transitional pathways are all fundamentally social, political, and relational rather than technical, and successful transitions consistently require multiple reinforcing pathways operating simultaneously; most theoretically consequential, the integrated analysis confirms a justice-viability nexus in which justice and viability are mutually constitutive, with every major vulnerability corresponding to a specific justice failure and every major pathway operating through a specific justice mechanism; and the rate of new thematic content declined sharply by 2025, suggesting that within the five-year period studied, the series had converged on a stable core conceptual framework for understanding SSF transitions, a pattern this thesis interprets as a feature of the period analyzed rather than a claim about the long-term trajectory of an ongoing knowledge platform. This thesis advances three principal theoretical contributions: the V2V Transition Ecosystem, an original tri-layered framework integrating the vulnerability-to-viability axis, the blue justice lens, and social-ecological systems theory; an original typology of five SSF vulnerability-viability transition contexts mapping specific vulnerability configurations to effective pathway combinations; and the justice-viability nexus, demonstrating that justice and viability are mutually constitutive rather than separate objectives - with direct implications for SSF policymakers, development organizations, community organizations, and international governance bodies seeking to support just and durable SSF transitions from vulnerability to viability.
  • Thumbnail Image
    Item type: Item ,
    The Cognitive-Affective Model of Sexual Consent Communication
    (University of Waterloo, 2026-08-04) Edwards, Jessica
    Sexual consent communication involves direct, unambiguous communication about willingness to participate in sexual encounters; promoting this form of communication is important to reducing sexual coercion and promoting the quality of sexual relationships. The present studies introduce and evaluate the Cognitive-Affective Model of Sexual Consent Communication, which is proposed to identify intrapersonal precursors (i.e., attachment insecurity) and mechanisms (i.e., perspective-taking, emotion dysregulation, experiential avoidance) that influence how adequately people communicate sexual consent. The model also indicates that these factors predict sexual consent motivations, an important precursor to the decision to enact affirmative consent behaviours. In Study 1 (N = 153), I examined the associations between core constructs of the model for people in current sexual relationships, finding that attachment, perspective-taking skill, and emotion regulation skill are associated with consent motivations and consent communication behaviours in existing sexual relationships. Study 2 included the piloting of a novel, vignette-based measure of sexual consent communication behaviour (Study 2a; N = 97), and testing of the key mediational pathways in a sample of single adults with this measure (Study 2b; N = 173). Results indicated support for the cognitive pathway of the model, but did not support the role of experiential avoidance in predicting consent motivations/behaviour. Finally, Study 3 (N = 212) tested the effects of a brief psychoeducational perspective-taking intervention on participants’ sexual consent motivations and behaviours. Outcome variables were assessed using the same vignette measure as Study 2, but participant responses were audio-recorded to extract more information about the emotional tone of consent communication. The intervention did not produce changes in consent communication behaviour relative to control, but sexual consent motivations were higher following the intervention for the subset of participants who reported low attachment avoidance. Taken together, the results of these studies illustrate the importance of moving beyond skills instruction to incorporate psychological traits and skills in consent education. Furthermore, the current research sheds light on the potential benefits of reducing perceived emotional threat in consent communication while enhancing perceptions of consent as rewarding and beneficial.
  • Thumbnail Image
    Item type: Item ,
    Lanternfish: Better random networks through optics
    (University of Waterloo, 2015-07-23) Szepesi, Tyler; Wond, Bernard; Kazhami, Fiodar; Rizvi, Sajjad; Brecht, Tim
    Current random datacenter network designs, such as Jellyfish, directly wire top-of-rack switches together randomly, which is difficult even when using best-practice cable management techniques. Moreover, these are static designs that cannot adapt to changing workloads. In this paper, we introduce Lanternfish, a new approach to building random datacenter networks using an optical ring that significantly reduces wiring complexity and provides the opportunity for reconfigurability. Unlike previous optical ring designs, Lanternfish does not require wavelength planning because it is specifically designed to provide random connectivity between switches. This design choice further reduces the difficulty of deploying a Lanternfish network, making random datacenter networks more practical. Our experimental results using both simulations and network emulation show that Lanternfish can effectively provide the same network properties as Jellyfish. Additionally, we demonstrate that by replacing passive optical components with active optical components at each switch, we can dynamically reconfigure the network topology to better suit the workload while remaining cost competitive. Lanternfish is able to construct workload specific topologies that provide as much as half the average pathlength than a Jellyfish deployment with twice as many switch-to-switch connections.