26734 research outputs found
Sort by
“We Turned Pain into a School of Life.” Rwandan Young People as Archivists of Knowledge in Response to Childhood Experiences of Intimate Partner Violence
No full textIntimate partner violence (IPV) remains a major public health and human rights issue worldwide, with serious consequences for both adult survivors and children. In Rwanda, where IPV prevalence remains high, national data often focus on adult experiences, with limited attention to how young people understand and respond to IPV between their parents or caregivers. Research on childhood experiences of IPV (CEIPV) tends to emphasize psychological harm and trauma over coping responses and agency, often portraying young people as passive witnesses. Even the emerging scholarship that resists this framing is mostly based in Global North contexts. This dissertation challenges these limitations by foregrounding Rwandan young people (YP) with CEIPV as storytellers and archivists of knowledge in ways that are important to them. Guided by Ubuntu ontology, an extended epistemology, critical pedagogy, storytelling, and resilience theories, the study engaged 13 Rwandan YP (aged 22–25) with lived experiences of CEIPV to understand how they made sense of their experiences. Through a participatory digital archiving methodology, the study explored how they name, story, and historicize their responses, as well as the benefits and limitations of sharing their CEIPV stories. Over three months, participants co-created a multimodal archive utilizing textual, oral, visual, and digital formats. Results show that participants resisted dominant trauma narratives through multistoried, sensory, and culturally anchored expressions. Their contributions highlighted harm, but also survival, care, response-ability, hope for restoring cycles of safety and the generational transmission of skills of resistance, particularly from mothers. Many drew on embodied memory and local metaphors to craft stories that honoured both vulnerability and strength. The archive became more than just a collection of experiences; it functioned as a living repository of mutual learning, resistance, and advocacy. Group sessions supported collective reflection, co-analysis, and discussion on ethics, representation, and the risks and possibilities of sharing stories in digital form. This work contributes to research, practice and policy on CEIPV, participatory and multisensory methods, and youth-centered knowledge systems. It offers a model of collaborative storytelling, story witnessing and story preservation that amplifies YP perspectives and affirms their role as active agents in shaping more just futures
A psycholinguistic approach toward an instance of a language change: The case of -esh in Persian
No full textIn the spoken dialect of Tehrani in Persian, a new marker -esh optionally attaches to 3SG past verbs indexing the subject. While there is a claim that this marker has been borrowed from the Persian object clitic paradigm and grammaticalized as a new subject agreement marker to fill the gap in the agreement paradigm (Rasekh, 2011, 2017), it still has some properties different from other agreement markers such as being optional on verbs and blocking subsequent object clitics –two properties which point to its clitic origin. Through a set of experiments, this thesis is aimed at exploring how this marker is processed by native speakers and how results obtained from real-time implicit tasks (i.e., self-paced listening tasks and visual-world eye-tracking paradigm) are different from offline explicit tasks (i.e., acceptability judgement task and forced-choice task). The results of real-time implicit tasks show that (i) speakers process the new marker -esh as a grammatical marker and different from its object clitic counterpart, and (ii) in the ambiguous condition (i.e., when both subject and object are 3SG.SPEC), the new marker -esh is processed as an object clitic. The results of offline explicit tasks show that (i) -esh can equally index specific and non-specific subjects, which points to its agreement property because clitics must index only specific arguments in Persian but not to the extent that speakers prefer the new marker -esh over the standard version where no overt agreement marker attaches to the 3SG past verbs. These results confirm the initial hypothesis that the new marker -esh is a new subject agreement marker, but the process of grammaticalization is not complete yet and it has still some properties of its clitic counterpart and speakers still prefer to see the standard version of 3SG where there is no agreement when asked explicitly
Engineering and Validating Soft Sensors for Interactive and Biomedical Applications
No full textSoft materials such as foams and textiles offer exciting opportunities for creating interactive devices that seamlessly integrate electronics into wearable, comfortable, and deformable forms. However, imparting elec-tronic functionality into these materials while preserving their inherent softness, breathability, and flexibility remains a significant challenge, especially when aiming for precise device patterning using accessible, low-cost fabrication techniques. This thesis addresses these challenges by introducing two novel platforms, FabFoam and rGOSense, that enable multi-functional sensing and interactive applications on soft substrates. FabFoam presents a scalable stencil-printing approach to pattern highly viscous functional inks directly onto foam substrates. This technique preserves the foam’s softness and elasticity while enabling the cre-ation of precise electrode designs for deformation sensing, touch input, electrotactile haptics, and on-body biochemical sensing. FabFoam’s sensors achieved signal-to-noise ratios (SNR) of approximately 34 for touch sensing and 25 for deformation across all inks, demonstrating reliable performance for interactive applica-tions. Electrochemical characterization showed that FabFoam’s sweat glucose sensor achieved a sensitivity of 2.74 μA/μM and a detection limit of 4.67 μM, confirming its potential for non-invasive glucose monitoring in wearable formats. Complementing this, rGOSense introduces a reduced graphene oxide (rGO)-coated nonwoven polyester fabric platform that offers superhydrophilic surface properties while retaining the textile’s softness and breathability. The rGO-coated textile demonstrates surface conductivity of 1.7 × 104 Ω/sq, which is about 2–10 times lower resistivity than several previously reported graphene- or carbon-based conductive textiles, enhancing electrical performance without compromising comfort. The rGOSense platform supports diverse sensing modalities, achieving SNR values of around 34 for touch sensing and approximately 26 for deformation sensing across multiple trials. Furthermore, its sweat glucose sensors achieved a sensitivity of 0.12 μA/μM
and a detection limit of 0.47 μM, highlighting improved electrochemical performance suitable for practical wearable biosensing. Collectively, this thesis establishes accessible, low-cost fabrication workflows for integrating advanced sensing capabilities into soft materials, enabling new possibilities in wearable health monitoring, ubiquitous sensing, and human-computer interaction (HCI). Through systematic materials characterization, technical evaluations, and demonstrations of interactive scenarios, this work contributes significantly to the field of soft interactive devices by bridging material science innovations with practical HCI applications
Coyote (Canis latrans) Behavioural Adaptations Amid Seasonal Weather Changes
No full textAs urbanization overtakes natural habitats, understanding its effects on wildlife behaviour is crucial for the welfare of both human and non-human animals. Adaptation to urban landscapes can mean survival or extinction for some species. A key transformation from urbanization is the Urban Heat Island (UHI) effect, where surfaces like pavement and rooftops raise city temperatures above those of surrounding areas. Coyotes (Canis latrans), known for their resilience, may serve as ecological indicators of broader impacts on less adaptable species. In particular, pup play – an ‘energetically costly’ behaviour tied to development and survival – may be sensitive to rising temperatures. I hypothesized that UHI-related temperature increases impacted play behaviour. Using non-invasive camera traps, I observed three post-natal homesites along an urban-to-rural gradient in Calgary, Alberta. I developed activity budgets (active: travelling, eating; play: chasing, stalking) and linked daily behaviour percentages to land surface temperatures (NASA MODIS data). I tested for behavioural differences between sites and examined whether urban pups played more at cooler periods, like nighttime, compared to peri-urban and rural pups, when the day temperature was high. Urban coyotes spent significantly less time in active (non-play) behaviours than those at Peri-Urban and Rural sites, while urban pup play occurred at significantly higher proportions. Temperature readings varied by source and site. At the Peri-Urban and Rural sites, camera temperatures differed significantly from weather station data likely due to vegetation or canopy cover. Urban site readings were more uniform. Satellite land surface temperatures differed significantly from ground-based measures across all sites raising concerns for ecological studies relying solely on remote sensing. Temperature did not significantly predict proportion of time spent in play behaviours, though a slight negative trend appeared at the Urban site. Play increased over time at the Peri-Urban site, possibly reflecting seasonal or developmental patterns. Nighttime play was not significantly influenced by temperature or site. These findings inform mitigation efforts, a deeper understanding of how resilient species like coyotes can adapt to changing environments, and the findings may serve as the 'canary in coalmine' for other species effects
The Weight of What Once Was
No full textThread by Thread is a sculptural installation and durational performance that represents the generational impacts of trauma. Drawing from trauma theory, affect studies, and autoethnographic methodology, the work addresses how unprocessed childhood abuse shapes identity and contributes to the formation of negative coping mechanisms such as substance use disorder. The installation consists of cement-soaked children’s clothing suspended on clotheslines. These clothes, once soft and symbolic of growth, are made rigid, heavily burdened, and fixed in time. In a durational performance, I slowly and deliberately attempt to clean these garments one at a time. By doing so, the piece re-enacts the lengthy emotional labour it takes to heal from trauma. This research positions my artistic practice as both a testimony and therapeutic tool. It resists the use of explicit narrative and instead creates an affective atmosphere where the viewer is invited to personally reflect and empathize. Material choice is central to the work’s meaning, with each material chosen for specific reasons associated with domesticity and mundanity. For example, children’s clothing is used because of its association with childhood developmental years. Specifically, Thread by Thread contributes to contemporary conversations on how art can make visible the burdens of trauma, address substance addiction, challenge stigma, and offer space for both individual and collective understanding. It highlights the power of installation and performance to represent traumatic testimony, foster empathy, and serve as a mechanism for repair. It creates an intersection between trauma studies, autobiographical practice, materiality, and affective engagement that expands the possibilities of art being used as a tool to evoke healing and social transformation
Surrogate-Guided Classification via Mixture Modeling
No full textIn many real-world classification tasks, obtaining accurate labels is costly or infeasible, and practitioners often rely on imperfect surrogates such as noisy human annotators. This thesis proposes a surrogate-guided classification framework that integrates covariate information, surrogate labels, and a structurally informative auxiliary variable through a probabilistic mixture modeling approach. The method flexibly accommodates both non-differential and differential misclassification, where the annotator error may depend on the true class label as well as covariates. An auxiliary variable, derived from or jointly influenced by the covariates and true label, is explicitly modeled to enhance the identifiability and accuracy of the latent class structure. Estimation is carried out via the Expectation-Maximization (EM) algorithm, leveraging the latent nature of the true class label. Through comprehensive simulation studies and a real-data application, the proposed method demonstrates substantial gains in classification performance, especially in settings with limited or unreliable annotation, highlighting the effectiveness of using surrogate information in semi-supervised learning contexts
Prairie Partnerships: Weaving Hydraulic Modelling and Traditional Knowledge in a Community- Led Framework
No full textHydroelectric dams on the Canadian Prairies fragment cultures and environments. Community X is a First Nation downstream of one such dam built in the 1960’s; the effects of which have not been investigated with considerations to their traditional knowledge or priorities. Community X has a long oral tradition of the area predating the regulatory structures and policy in the area, as well as environmental records such as flow data. Therefore, they have a long and continuous dataset of landscape changes within their stories that is central in understanding the environment of the region, and water dynamics in a flat landscape. This research aims to assess the modern and historic surface water dynamics in and around the community utilizing a transdisciplinary, community-led approach. To accomplish this, a two-dimensional hydraulic model was developed in HEC-RAS. This model was then refined through community engagement with Elders and guides from the Nation. Community engagement sought to catalog direct and indirect anthropogenic impact to Community X and their traditional territory. UAV-based LiDAR was used to better understand the topography of key areas to support the modeling. To better understand the historic context surrounding the construction of the dam, historic air photos taken before and after construction were used alongside interviews with Elders. Air photos were collected from the SANDS library at UCalgary. Air photos, model results, interviews, and drone imagery were then uploaded into an interactive map for the use of the Nation. The goals of this research are to understand the changing environment due to human activities, and to create a useful tool for the Nation's leadership for planning purposes
Chatbotting about the topics of CORE 205 Introduction to disability studies: A teaching tool
No full textThis document is myself chatbotting with Chat GPT, Kimi2 and Claude AI about the topics I will teach in the fall of 2025 in my CORE 205 Introduction to Disability Studies class. More in the section why I did generate the output. In short the idea is that students can compare the chatbot results with the course material, which, for example, includes “Teaching about the intersectionality of disabled people using an intersectional pedagogy framework: a primer” [28] and other material specific to a given week and the lectures
A Secure and Privacy-Preserving, Meta-Learning-Based Machine Unlearning Approach Using Redactable Blockchain under Intent-Based Networking Orchestration
No full textIntent-Based Networking (IBN) is an emerging paradigm that enables automated, policy-driven management of systems by translating high-level intents into behavior. As IBN increasingly relies on Artificial Intelligence (AI) for adaptive decision-making, concerns around data privacy and regulatory compliance become critical. Specifically, the Right to Be Forgotten (RTBF), as defined by the General Data Protection Regulation (GDPR), mandates the removal of both personal data and its learned influence from AI models. Traditional unlearning methods that retrain models from scratch are computationally costly and unsuitable for infrastructures. To address this, this thesis proposes a secure and privacy-preserving machine unlearning framework that leverages meta-learning to remove learned data influence without full retraining. Redactable blockchain ensures auditable data handling, while Secure Multi-Party Computation (MPC) enables collaborative and trustworthy redaction. Orchestrated by IBN, this framework aligns AI model behavior with privacy intents, enabling scalable and regulation-compliant unlearning
Effect of Metallurgical Structure on Hydrogen Diffusion and Distribution in Steel
No full textHydrogen has been regarded as an ideal alternative to fossil fuels to achieve the 2050 net-zero goal while meeting the rapidly increasing energy demand. Transporting large amounts of hydrogen via pipelines is the most effective and economical methods. However, hydrogen embrittlement (HE) may occur to endanger the structural integrity and operational safety of the pipeline. Hydrogen induced pipeline degradation is strongly dependent on the trapping behavior of the steel lattice structure. Various microstructures often found in pipeline steel consist of features which serve as potential traps for hydrogen. These trap sites include dislocations, grain boundaries, alloying elements, defects and interfaces. Thus, it is important to study the effect of metallurgical structure on hydrogen diffusion and distribution in steel. In this work, we first used scanning Kelvin probe force microscopy (SKPFM) measurements and finite element modelling to study the diffusion and distribution of hydrogen atoms at metallurgical microphases contained in X52 pipeline steel. Results show that the pearlite contained in the steel is more stable than the ferrite during electropolishing, as indicated by the measured topographic profiles and contact potential differences (CPD). The hydrogen charging enhances the electrochemical activity of both pearlite and ferrite, as shown by increased contact potential differences and thus the decreased work function. As the hydrogen charging time increases, the contact potential differences of both phases further increase, implying that their activities increase with the H-charging time. The pearlite has a greater contact potential difference and thus a lower work function than the ferrite. This is associated with more H atoms accumulating at the pearlite than at the ferrite. The H atom diffusion and accumulation are affected by H diffusivity at phase boundaries, H-trap binding energy and the number of traps in the steel. Then, considering that some macroscopic defects such as dents, corrosion pits, and notches are found on X52 pipeline steel during its long-term service, causing local stress concentration, we investigated the distribution of hydrogen atoms at a notch on an X52 pipeline steel under various applied loads or displacements by using a scanning Kelvin probe and finite element modeling. Results show that when the stress concentration at the notch is elastic, the hydrogen atom diffusion and distribution are driven by stress, rather than strain, in the steel. However, when displacements are applied to the steel specimen, the strain dominates the H atom distribution. As the local stress or strain increases, the H atom concentration increases, as indicated by negative shifts of the contact potential difference. For newly constructed hydrogen transportation pipelines, it has been reported that using higher grade steels for large-scale, long-distance hydrogen transportation can effectively reduce overall costs. Thus, we analyzed the effect of metallurgical microphases on H-atom accumulation in the X80 steel. Results show that hydrogen charging elevates the electrochemical dissolution activity of both ferrite and bainite contained in X80 steel, as indicated by the increased contact potential difference and thus the decreased work function. The H atoms tend to accumulate at ferrite, making the local hydrogen atom concentration much greater than the concentration at bainite. This phenomenon may partially result from the mechanical polishing process applied during sample preparation. Given that ferrite is softer than bainite, it is more susceptible to the generation of a high density of dislocations during polishing. These dislocations can act as potential hydrogen trapping sites, promoting the accumulation of hydrogen in ferrite. This possible effect of polishing has now been noted to avoid any misunderstanding regarding the hydrogen distribution in X80 steel. The results imply that, compared with bainite, ferrite is the location to accumulate more hydrogen atoms initiating hydrogen-induced cracks once the local hydrogen atom concentration reaches a threshold value, particularly under external loading conditions during service. Furthermore, considering the new pipelines may be constructed in different environmental conditions, we used nanoindentation tests, combined with hydrogen permeation test to study the effect of solution pH and current density on H permeating parameters and nanomechanical properties of X80 pipeline steel. Results show that the steady-state hydrogen permeating current density, hydrogen permeating flux, and the hydrogen subsurface concentration obtained in 0.5 M H2SO4 solution are apparently greater than the parameters obtained in near-neutral pH NS4 and 0.1 M NaOH solutions. The changes in nanoindentation of the steel subjected to electrochemical hydrogen charging depend on the charging solution used. In the 0.5 M H2SO4 solution, as the electrochemical H-charging time increases, the nanoindentation of the steel continues to decrease. However, in NS4 and 0.1 M NaOH solutions, the steel nanoindentation increases with time. This phenomenon is attributed to the formation of corrosion product films on the steel surface. The softening and hardening transition phenomenon occurs on the steel subjected to electrochemical hydrogen charging. Although measurement errors may cause variations in the nanoindentation results, the evolution of nanoindentation upon electrochemical H-charging exhibits a discernible tendency. H atoms could affect the elastic modulus of steel. When electrochemical H-charging promotes a uniform distribution of H atoms within the steel, achieved through either extended charging time or application of a cathodic current density, the elastic modulus of the steel tends to increase