82837 research outputs found
Sort by
Insight into government, November 21, 2025
No full textAlberta's independent newsletter on government & politics.
Issue has a typo saying it is the November 14, 2025, but it is the November 21, 2025 issue
Digital Literacy Practices and Learning at Home for Children of First Generation Newcomers in Canada
No full textContemporary childhood exists in a rapidly changing literacy context in the digital age, where digital devices and technology are increasingly used at home and in school. This qualitative study explored newcomer children’s digital practices, as reported by their parents, and parents’ perspectives of their children’s digital usage at home. The COVID-19 pandemic and shifts to online remote learning have also impacted children’s digital literacy practices, and this study also provides understanding of culturally and linguistically diverse children’s digital literacy practices at home during this pandemic era.
Theoretically, this research is framed by contemporary literacy studies, specifically, multiliteracies and new literacy studies, which understand literacy as everyday multimodal and multilingual social practices, across multimedia. I drew on Kress’s (1997, 2005) concept of multimodality in the digital context and Green’s (1988, 2012) three-dimensional literacy model as my conceptual frameworks to help me understand and interpret my data on children’s digital practices and how parents understand their children’s digital literacy practices at home.
My study is a qualitative case study using ethnographically informed approaches and small stories to present the findings. The data collection tools I used were an online survey and semistructured interviews with first-generation newcomer parents to answer my research questions. I received 39 responses to the online survey and interviewed 4 first-generation newcomer parents. In this dissertation, I focus on the interview data with 3 of the newcomer parents. The findings indicate that the three children’s digital practices at home are situated in a contemporary context where digital devices and digital environments are essential for their social networking, family relationship building, learning, entertainment, creation activities, and recording life moments through photos and videos. My research shows that children are not only consumers of digital texts but also active producers. The three children have been exposed to social media influencers, particularly on the YouTube platform. Today, children are not always the only literacy learners at home and the parents are not always the literacy knowledge holders, as was typical in literacy environments historically. Instead, within the digital multimodal literacy context, children may know more than their parents and assist their parents’ technology usage at home. In terms of parents’ perspectives towards their children’s digital practices at home, the three participating parents tended to focus on their children’s operational skills with digital devices in the interviews and sometimes provided examples of how their children engaged in digital activities by following conventions and rules. However, the three parents did not share much about how their children use critical thinking and analysis in their digital practices. In addition, the parents described the difficulties of balancing their children’s digital usage at home in this contemporary era.
The insights gained from my study have the potential to provide valuable information for newcomer parents, educators, and policymakers to understand newcomer children’s home digital literacy experiences. As well, this information may ease the anxieties of newcomer parents about parental management of digital practices and digital literacy. Educators and policymakers will be able to use this study to recognize the ways that newcomer children use digital tools in their homes as they consider curriculum, pedagogy, and policy for newcomer children
De Novo Synthesis of Bicyclic Nucleoside Analogues
No full textNucleoside analogues are synthetic forms of nucleosides and contain a nucleobase and sugar moiety. Nucleoside analogues are used in the pharmaceutical industry as they have a vital impact in treating many diseases, particularly as small molecule therapeutics. Nucleoside analogues often require lengthy sequences to synthesize due to the use of natural sugars in which many protecting and deprotecting steps are utilisied.
This thesis aims to establish a [4.3.0]- bicyclo nucleoside analogue, an unexplored class of nucleoside analogues, in a short, concise sequence using a de novo approach. Progress towards a more modular and atom-economical synthesis is discussed. It is worth noting that [4.3.0]-bicyclo nucleoside analogues are not heavily studied in RNA or DNA therapy and, as such, may provide new opportunities in the development of oligonucleotide therapeutics.
The synthesis of the natural product Malayamycin A is presented in Chapter 3. Malayamycin A possesses a bicyclic nucleoside core and was previously synthesized by Hanessian in 25 steps. The proposed route in this thesis utilizes a de novo strategy and aims to synthesize the target molecule in ~12 steps
Data Augmentation, Explainable AI, and Conversational Diagnosis via Large Language Models: A Path Towards a Real Automated Clinical Mental Disorder Diagnosis
No full textMental health disorders—particularly Major Depressive Disorders (MDD) and Post-Traumatic Stress Disorder (PTSD)—pose significant challenges for accurate and timely diagnosis, given the subjective nature of clinical evaluations, the scarcity of labeled data, and the heterogeneous presentations of symptoms. This thesis aims to advance computational psychiatry by integrating large language models (LLMs) and deep-learning techniques to address three critical obstacles: (1) data scarcity and imbalance, (2) the opacity of “black-box” diagnostic models, and (3) the need for adaptive, proactive reasoning in complex diagnostic scenarios.
First, we tackle the data challenge through LLM-based synthetic data generation. By designing zero-shot and few-shot text augmentation strategies, we create mock clinical interview transcripts that closely emulate real patient narratives, thereby mitigating small or skewed datasets without incurring extensive data collection costs. Empirical results demonstrate that these synthetic transcripts significantly enhance model robustness and reduce overfitting in automated PTSD diagnosis.
Second, we address the explainability gap by developing a dual-layer framework that merges system-level feature analysis (e.g., attention mechanisms, Shapley values) with human-level “causability.” Leveraging retrieval-augmented LLMs, our pipeline translates raw, model-centric explanations into clinically interpretable narratives aligned with established diagnostic criteria. This synergy between system-level and human-level interpretability fosters greater clinician trust and lays the groundwork for regulatory acceptance of AI-driven mental health diagnostics.
Third, we propose a proactive conversational AI system, termed WiseMind, that mimics the iterative, multi-step reasoning processes of experienced psychiatrists. Through a multi-agent design, WiseMind adapts questions based on the patient’s evolving context, supports differential diagnoses, and refines its line of inquiry in real time. This paradigm shift from static classification to dialogue-based assessment better captures the nuanced decision-making required in clinical practice.
Finally, we introduce two complementary frameworks—Reasoning-Aware Self-Consistency (RASC) and Derailer-Rerailer—to optimize LLM inference costs while maintaining or improving diagnostic accuracy. By selectively refining or verifying outputs only when the model exhibits inconsistent or unstable reasoning, these methods achieve substantial reductions in computational overhead.
Evaluations across multiple datasets, including user studies and clinical expert assessments, confirm that the proposed solutions significantly enhance diagnostic accuracy, interpretability, and computational efficiency. Taken together, these contributions mark a step toward AI-driven mental health assessments that are not only more robust and scalable but also transparent and clinically aligned, thereby opening new avenues for data-centric, explainable, and proactive AI in psychiatric care
Diamonds as probes of earth's mantle in space and time
No full textDiamonds probe the otherwise inaccessible interior of Earth since at least ~3.5 Ga. In this thesis, two diamond populations from the Canadian portion of the Laurentian supercraton are utilized to investigate the composition and evolution of Earth’s mantle through a wide range in time and depth.
In Chapter II, a batch of more than 3000 diamonds from ~2.73 Ga ultramafic volcaniclastic rocks in the Oxford Lake - Knee Lake greenstone belt provides information on the thermal state of the lithospheric mantle underlying the northwestern Superior Craton during its accretionary period. The Knee Lake diamonds show unusual physical characteristics relative to typical kimberlite-hosted diamonds, which, together with a correlation between crystal shape and nitrogen aggregation, point to a single, rapid growth event in a cool mantle substrate driven by strong C-oversaturation of the growth medium. Pressure-temperature estimates from a garnet-olivine inclusion pair indicate diamond formation at conditions equivalent to a cool, modern-day reference geotherm of 38–39 mW/m2. At ~2.73 Ga the lithospheric column also contained eclogite formed by metamorphism of a recycled crustal protolith, as indicated by an omphacite inclusion and low δ13C values of the host diamond and other diamonds of the suite. The Knee Lake diamond suite likely records cool transient thermal conditions inherited from subduction stacking of lithospheric slabs to form the lithospheric mantle of the western Superior Craton.
In Chapter III, secondary ion mass spectrometry analyses of 186 Knee Lake diamonds yielded the lowest (-46.8‰) to the highest (+31.8‰) δ15N values ever measured for terrestrial diamonds. The extreme range and discontinuous distribution of the δ15N values rule out an explanation through mixing of heterogeneous N sources and its origin remains enigmatic. The systematics of N concentration and δ15N variations for most of the diamonds are permissive of a Rayleigh distillation process. The very large δ15N range, that is uncorrelated with δ13C or N concentration, excludes equilibrium fractionation processes. Kinetic isotopic effects associated with decomposition of N-phases could produce the most 15N-enriched compositions, but not the most 15N-depleted. The extreme δ15N values of the Knee Lake diamond suite are correlated to crystal shape, with the most negative values occurring in cuboid and the most positive in octahedral crystals. This suggests the operation of fractionation processes related to different growth-surface structures or growth conditions. Distinct δ15N values in different growth sectors of mixed habit synthetic diamonds are confirmed by the analysis of seven synthetics. However, ten Knee Lake natural diamonds of mixed habit growth show only small and non-systematic internal δ15N variability related to growth sectors. Models of N isotope diffusive fractionation along thermal and chemical gradients are explored. Thermal diffusion cannot explain the large δ15N range of the Knee Lake diamonds and would predict opposite fractionation directions. Diffusion along chemical gradients is the only process that can produce the low δ15N values. These observations call for the operation of various kinetic effects during diamond growth, likely superimposed on source variability.
Chapter IV focusses on diamonds from the Candle Lake C29/30 kimberlite, erupted on the northwestern edge of the Sask Craton, Canada. The characteristics of the Candle Lake diamonds suggest a complex population, composed of at least two subgroups. The larger diamonds are more often highly resorbed, internally pure and N-rich, and derive from lithospheric eclogites as substrates, as indicated by their inclusion mineralogy and δ13C values. The N characteristics of the lithospheric diamonds at Candle Lake are consistent with a diamond formation age of ~1.27 Ga reported for the Fort à La Corne deposit, suggesting a widespread Proterozoic diamond growth event in the Sask Craton. Diamonds with pristine octahedral shape, often aggregates of multiple crystals, are commonly of smaller dimensions and N-free. They have C-isotope compositions close to the mantle value and carry sublithospheric mineral inclusions of metaperidotitic paragenesis. Inclusion assemblages and residual pressures attest to their derivation from lower mantle to transition zone depths. The occurrence of carbonate inclusions in this diamond suite suggests diamond formation from carbonatitic melts at oxygen fugacity higher than in ambient mantle. The pristine shape of the sublithospheric diamonds may be explained by their upward transport to lithospheric depth only after the intense resorption event affecting the lithospheric diamonds
Real-Time Digital-Twin for Synergistic Interaction of SMRs and Sustainable Power Systems
No full textAmid the challenges of digitalization, coordination complexities, and uncertainty in integrating diverse energy sources, digital-twin technology and small modular reactors (SMRs) are increasingly emerging as cutting-edge tools and technical pathways to address these issues. Against this backdrop, this article proposes a modular, scalable real-time digital-twin (RTDT) and surrogate physical-twin (SPT) collaborative hardware framework that predicts and responds to anomalies six times faster-than-real- time, while also addressing simulation interface issues across multiple physics modalities. The proposed system, deployed on a multi-embedded field-programmable gate array platform, facilitates communication via the Aurora 64b/66b protocol and gigabit Ethernet based on IEEE 802.3, evaluating the impact of latency and data integrity on real-time performance. The SPT case study explores interactions among SMRs, the IEEE 39-bus system, and the CIGRÉ B4 DC grid with wind farms. The RTDT-SPT platform highlights the advantages of integrating next-generation advanced nuclear technology, renewable energy, and digital management systems by enhancing the coordination of SMRs in primary and secondary frequency control, reactor-leading mode, and turbine-leading modes, effectively smoothing the volatility of wind power generation
Indigenous Women and 2SLGBTQIA+ Perspectives on Chronic Wasting Disease Research and Governance: #cwdbeadsproject and Indigenous STS Interventions
No full textAwareness of chronic wasting disease (CWD)--a fatal neurodegenerative disease primarily affecting cervids (deer, moose, elk, and caribou)--and communication of important information regarding spread, transmission, and risk associated with the disease are the responsibility of provincial and territorial governments. Provincial messaging available to the public regarding CWD focuses on recreational hunter populations in Alberta, whereas Indigenous harvesters remain increasingly uninformed.
The purpose of this research was to explore the perspectives of Indigenous beaders--a group that interacts with cervids as part of the resurgent practice of harvest, hide tanning, and beadwork, but that remains consistently not consulted on matters of wildlife management and health. The objectives of this research were to provide current and concise information on CWD to the beadwork community, to engage Indigenous women and 2SLGBTQIA+ people on CWD, and to include the perspectives of Indigenous beaders (Indigenous women and 2SLGBTQIA+ people) and their other-than-human counterparts into research on CWD management. The following three research questions guided the investigation of this project: 1) What concerns do Indigenous beaders offer regarding CWD? 2) How does Indigenous beadwork function as a co-constitutive process between human and other-than-human relations? And 3) How do Indigenous beaders imagine Indigenous communities being involved in research and management of CWD? To answer these questions, this research used a qualitative approach that takes direction from critical Indigenous methodologies and arts-based research creation. Data was gathered using beadwork elicitation via the #cwdbeadsproject, and conversational interviews. In total, 17 Indigenous beaders participated in this research, and a total of 21 combined interviews and submissions collected. This research finds that Indigenous beaders are engaging in harvest related practices in ways that are not visible to provincial governments currently managing CWD. Subsequently, the lack of engagement with beaders results in knowledge gaps where beaders are less aware of and do not have access to information or safe handling or processing protocols that can greatly assist them in reducing the possibility of disease transmission and spread.
Further, this research sought to better understand the co-constitutive relationships between Indigenous beaders and their other-than-human counterparts. In this regard, beaders communicated a level of respect akin to care that is in alignment with a respect for animals and their sacrifice in the hunt. This care was expressed through the idea of using the animal's body to its fullest capacity. The final results chapter presents Indigenous beaders perspectives on Indigenous-led initiatives and community involvement in CWD, insight into the creative potential of Indigenous science, and a series of future research objectives for critical Indigenous scholars looking to engage in Indigenous scientific projects regarding wildlife disease. In summary, this research provides theoretical contributions to the emerging field of Indigenous Science, Technology, and Society (STS) by incorporating critical perspectives on the technologization of wildlife disease management and the exclusion of Indigenous expertise on wildlife health. Further, this research contributes to Indigenous methodological approaches through the development of Indigenous multispecies autoethnography and arts-based research creation
Using Discrete Overlapping Partitions for Count-based Exploration
No full textExploration remains an open problem in reinforcement learning. Ideally, a useful exploration method should efficiently explore sparse reward environments, scale to large environments, and be simple to implement. Most agents use random-based methods like ε-greedy due to their simplicity and low computational cost. However, these methods struggle in sparse reward settings and take a long time to converge. Count-based methods encourage exploration in less-visited areas but do not scale well to large environments. Extensions to count-based methods to work with function approximation improve performance in complex environments like Montezuma's Revenge, but are rarely used due to their computational and implementation complexity.
We propose partition counting exploration, a new method that achieves all three desiderata simultaneously. Our exploration method handles large environments by maintaining counts within multiple overlapping partitions of states to derive exploration bonuses.
We evaluate our algorithms on three continuous observation environments where count-based methods cannot be applied, including MiniGrid DoorKey with image-based observations