HARVEST
University of Saskatchewan's Repository for Research, Scholarship, and Artistic Work
Welcome to HARVEST, the repository for research, scholarship, and artistic work created by the University of Saskatchewan community. Browse our collections below or find out more and submit your work.
Communities in HARVEST
Select a community to browse its collections.
Recent Submissions
Fostering FASD-Informed Practice with Incarcerated Adults with Fetal Alcohol Spectrum Disorder: Towards More Informed, Strengths-Based, Investigative and Connected Correctional Environments
(2024-06-21) Buttinger, Pam R; Dell, Colleen; Mela, Mansfield; Olver, Mark; Buhler, Sarah; Sharp, Maegan; Reid, Natasha; Ahmed, AG; Brooks, Carolyn
The capacity to foster an FASD-informed environment is predicated on the ability of those within the said environment to make their own practice and interactions FASD-informed. The aim of this study was to critically consider how Fetal Alcohol Spectrum Disorder (FASD)-informed practice could be fostered within Correctional Service of Canada (CSC) facilities. The study aimed to identify, first, the receptiveness of CSC staff towards adopting FASD-informed practices, and second, the essential elements of an FASD-informed environment and whether they could be applied to a correctional environment. To do so, the researcher carried out the research using a three-part, mixed-methods design that combined a traditional survey, an open-ended survey with a modified Delphi approach follow-up, and interviews with three separate professional groups. The groups included CSC frontline staff, professionals working within intervention and research for FASD, and CSC staff who work within an institution that also houses an FASD diagnostic clinic. The results indicated that FASD-informed environments employ practices and people who (a) are informed about the disorder, (b) are willing to employ strengths-based approaches, (c) are relentlessly investigative in their pursuit to help the individuals they work with find success, and (d) value the role of connection and relationships.
Within CSC, many staff reported being receptive to adjusting their approach when working with individuals with FASD; however, promoting widespread FASD-informed attitudes and approaches will require changes to current practice, necessitating further training and exposure. Mainstreaming FASD training requires an understanding that not all staff need the same level of FASD knowledge. A tiered implementation model (foundation, knowledge base, action, and specialized service) is offered as a guide to fostering FASD-informed correctional environments. At a foundational level, working towards all institutions being trauma-informed, person-centered, and strengths-based will be beneficial to all incarcerated individuals, not only those with FASD. This can be fostered by offering regular, multimodal training to all staff and fostering attitudes that promote interdisciplinary teamwork and safety. For those in specialized environments or disciplines, offering domain and discipline-specific intervention strategies and opportunities to practice them is suggested. Continued support of the FASD Diagnostic Clinic teams will also ensure continued diagnostic capacity, consultation, and specialized services. All four tiers in the tiered implementation model rely on strong managerial support and are strengthened through connections within the institution and the surrounding community. The implication of the findings within this study offers recommendations on how to respond to FASD within correctional systems that is consistent with CSC’s mandate to provide treatment aimed at rehabilitation and mitigating the risk offenders place on the safety of society. Given the overrepresentation of FASD within institutions, there is a need to provide employees with tools to work with this population and be responsive to these mandates. With top-level support to introduce sustainable training opportunities alongside recognizing that FASD requires frontline staff the room to be flexible and accommodating, a shift can begin towards fostering more FASD-informed correctional environments.
Comparing the vocabulary and grammatical structures of The Sims 4 with those of the Evolve 5
(2024-06-21) Habibi, Ashkan; Li, Zhi; Cenerini, Chantale; Koole, Marguerite
Using video games for language learning can be beneficial for language learners. While
previous studies have investigated the utility of video games for language learning, there are limited studies that compared the linguistic input in video games and that in existing language classes. This study delves into a comprehensive exploration of linguistic elements by comparing the vocabulary and grammatical structures present in the virtual realm of The Sims 4, a simulation video game, with those found in an English textbook called Evolve 5, published by Cambridge University Press targeting B2 or upper intermediate English language learners. The purpose of this study is to see whether the game has the potential to be used as an extramural activity outside the classroom along the book. The corpus of The Sims 4 script includes 323,555 words, and the Evolve 5 contains 294 targeted words or phrases, as featured in the textbook’s vocabulary sections and 20 grammatical structure groups distributed in 12 units.
In addressing the research questions of this study, two software tools were employed: LancsBox for identifying vocabulary and grammatical structure groups within the game script, and AntConc for visualizing the dispersion of words and phrases. The findings show that The Sims 4 is a valuable source for English language learning for both vocabulary and grammatical structures learning. The results show although The Sims 4 has a rich content for language learning, in terms of overall vocabulary coverage. However, the game seems to be most useful for teaching a few topics as only six out of 24 topics in the textbook show medium or higher coverage of vocabulary, that is over 45% of the targeted words or phrases in the corresponding topic appear six or more times in the game script. Regarding grammatical structures, it is an invaluable and appropriate source to be used as an extramural activity because out of 20 grammatical structures, 12 of them appear highly frequently in the script of the game (i.e., 40 or more occurrences). The findings of this research offer advantages to language learners and educators alike. By highlighting certain topics heavily featured in the game, educators can harness its potential to motivate learners to engage in it as an extracurricular activity beyond traditional classroom settings. Furthermore, the results emphasize the significance of incorporating digital gaming technologies into language education to promote immersive and engaging learning opportunities.
Epitope-TCR Interaction Prediction with Deep Learning based on Sequence and Physicochemical Properties
(2024-06-19) Raha, Rawshon; Wu, FangXiang; Liu, Qiang; McQuillan, Ian
Immune system cells are capable of defending our body from attack by a pathogen if they succeed to recognize the pathogen as a threat before its attack. The recognition of chewed up fragments of the antigen (epitope) by immune system cells (TCR) can be predicted by successful epitope-TCR recognition. However, testing numerous epitope-TCR sequences experimentally for interaction is very time and resource consuming. Predicting this interaction computationally before testing them in the laboratory can help with effective vaccination and personalized healthcare. In this study, I addressed the interaction prediction task in the unseen epitope setting by developing a pairwise combination based model, and in the unseen TCR setting by developing an ensemble learning model with sequence based calculations.
In the pairwise combination based model for unseen epitope-TCR interaction prediction, the pairwise epitope and TCR sequences are used simultaneously to generate images like features using absolute difference and vector outer product of constituent amino acid's physicochemical properties. The best performing physicochemical properties have been selected and found to exhibit much higher performance in comparison to the existing unseen epitope prediction models. The absolute difference based model produced an AUC of 0.64 with only two best performing physicochemical properties, namely, Hydrophobicity and Net Charge Index. The vector outer product based model produced an AUC of 0.60 with the same two properties. Furthermore, the model achieved an AUC of 0.82 by combining both types of features while the best competing model had an AUC of 0.55 for similar setting and dataset.
In the ensemble learning model for predicting unseen TCR-epitope interactions, the features were generated using physicochemical property vector, one hot vector, and ProtBERT embedding vector. During the model training, the equally-long sequences were created by zero padding and a masking strategy is adopted to mitigate the noises which may have been introduced by the zero padding. The best performing models using physicochemical property vector, one hot vector, and ProtBERT embedding vector achieved AUC values of 0.74, 0.78 and 0.77, respectively. Moreover, the ensemble learning model based on the individually predicted posterior probabilities achieved an AUC of 0.79, which is convincingly better than the existing best performing methods.
Biogeochemical Implications of Sulfate-Based Coagulants in Treated Oil Sands Fluid Fine Tailings
(2024-06-19) Adene, Philip A; Lindsay, Matthew; Birks, Jean; McBeth, Joyce; McPhedran, Kerry; Butler, Sam; Eglington, Bruce
Oil sands mining operations generate a large volume of fluid fine tailings (FFT), requiring incorporation into aquatic and terrestrial reclamation landforms. These tailings are a mixture of inorganic solids, oil sands processed-affected water (OSPW), and residual bitumen. The initial solid contents, which range between 25 to 35 % (w/w), are dominated by quartz and clay minerals (i.e., kaolinite, illite, and illite-smectite (I-S) mixed layer) with minor carbonates, trace sulfides, and oxides. Associated OSPW contains elevated concentrations of Na+, Cl-, HCO3-, and NH4+, plus trace elements, naphthenic acids (NAs), and residual hydrocarbons. Elevated Na+ concentrations relative to other cations (e.g., K+, Ca2+, and Mg2+) hinder aggregation of silt- and clay-size particles (≤ 44 µm), thereby slowing FFT settlement and dewatering, which precludes prompt integration into reclamation landscapes. Consequently, FFT inventories within tailings ponds have grown steadily over time, approaching 1.3 billion m3 by 2020. Regulators have issued directives intended to curtail these growing inventories and promote progressive mine reclamation. These directives, coupled with slow settlement behaviour, prompted oil sands operators to develop various technologies that accelerate dewatering with the addition of chemical coagulants. However, the influence of these coagulants on FFT biogeochemistry is not fully understood.
The main goal of this thesis was to examine the biogeochemical implications of chemical coagulant treatments on oil sands FFT and to understand the effects of cation exchange reactions associated with gypsum addition on the isotopic signature of FFT pore-water. This research uses the integration of analytical methods and anaerobic batch experiments to examine how different doses (i.e., 0, 500, 1000, and 1,500 ppm) of coagulants (i.e., alum [Al2(SO4)3.nH2O], ferric sulfate [Fe2(SO4)3.nH2O], and gypsum [CaSO4.2H2O]) affect FFT geochemistry, mineralogy, and microbiology over time. The research also uses Na-Ca exchange batch experiments with different masses (0.025, 0.05, 0.1, 0.2, 0.4, 0.8, 1.6, 2.4, and 3.2 g) of reference clay minerals (kaolinite, illite, and I-S) and FFT to determine the extent of Ca uptake during cation exchange reactions associated with gypsum amendments using the magnitude of isotopic fractionation.
The anaerobic batch experiments showed that both treated and untreated FFT samples host diverse microbial communities with a range of metabolic capabilities. Sulfate and Fe(III) reduction and methanogenesis are the key biogeochemical redox processes in FFT deposits. The interactions among these processes lead to the complex biogeochemistry of FFT. Sulfate reduction suppresses CH4 production, while the H2S produced reacts with microbially-derived Fe(II) to form Fe(II) sulfide minerals, which promote the sequestration of metals, which include As, Ni, V, and Zn. Coagulant addition enhances mass-transfer reactions (e.g., carbonate minerals dissolution-precipitation, ion exchange, adsorption-desorption, and sulfide minerals precipitation), which determines the release or attenuation of major and trace elements. The sulfate-based coagulants also decreased solution pH (alum: 7.16 to 4.50, ferric sulfate: 7.72 to 6.28, and gypsum: 8.03 to 7.41) and alkalinity (alum: 840 to 20 mg L-1, ferric sulfate: 920 to 280 mg L-1, and gypsum: 920 to 420 mg L -1) and increased electrical conductivity (EC) (alum: 3.78 to 9.06 mS cm-1, ferric sulfate: 2.31 to 10.71 mS cm-1, and gypsum: 5.98 to 10.30 mS cm-1) due to elevated Na+, Cl-, and SO42-. The Ca - Na exchange experiments revealed that cation exchange dominates Ca uptake in acidic environments, and the magnitude of isotopic fractionation depends on the extent of Ca uptake. Illite and I-S exhibited the most Ca uptake and isotopic fractionation compared to kaolinite. The clay minerals present in FFT control the Ca uptake and the magnitude of isotopic fractionation. Heavier isotopes (e.g., 44Ca) were preferentially attached to the clay surfaces in most samples, which could be controlled by variation in Ca2+ coordination number (CN).
Social Trust, Social Tolerance, and Board Gender Diversity in a Cross-Section of Countries
(2024-06-19) Kordi, Fatemeh; Maung, Min; Yang, Fan; Mishra, Dev R; Wynes, Mike; Tannous, George
The abstract of this item is unavailable due to an embargo.
Attachment insecurity and sleep disturbances: Exploring the association in individuals and their bedpartners, and the effects of co-sleeping behaviours
(2024-06-19) Castillo, Dyana; McWilliams, Lachlan; Symons, Doug; Tomczak, Corey; Gagnon, Michelle; Loehr, Janeen; Stacey, Adam
A growing body of research indicates that individual differences in attachment insecurity are related to sleep-related variables. In general, greater attachment insecurity (i.e., anxiety and/or discomfort in close relationships) has been found to be associated with sleep problems, such as pre-sleep hyperarousal (Palagini et al., 2018), and poorer sleep quality (Maunder et al., 2011). Research to date has primarily studied samples of individuals. When studying heterosexual couples, relationships between the variables from one type of couple member (e.g., male partners) are often referred to as “actor effects.” Research conducted with couples indicates that attachment insecurity is also associated with the sleep quality of relationship partners (i.e., “partner effects”). Further research exploring actor and partner effects involving adult attachment and sleep is warranted. The goal of Study 1 was to replicate actor and partner effects regarding adult attachment characteristics and ratings of sleep quality. In Study 2, the goal was to identify variables likely to be moderators (e.g., co-sleeping arrangements) and mediators (e.g., pre-sleep hyperarousal) of these associations. Study 1 utilized the actor-partner interdependence model to study attachment and sleep quality in a sample of 173 couples. It found actor and partner effects of attachment anxiety, but not avoidance. Study 2 involved a series of multiple regression analyses of attachment variables and confounds (age, depression, and health) on a variety of sleep-related variables in a community sample (N = 309). Attachment anxiety was a robust predictor of electronic device use in bed, dysfunctional beliefs about sleep, pre-sleep cognitive arousal, and pre-sleep arousing activities, as well as a global sleep outcome (i.e., daytime sleepiness). Attachment avoidance was a robust predictor of pre-sleep eating and drinking and pre-sleep cognitive arousal. Both attachment dimensions predicted the frequency of pleasant and neutral pre-sleep conversations with partners. Only one variable (dysfunctional beliefs about sleep) emerged as a mediator between attachment anxiety and a sleep outcome (i.e., daytime sleepiness). This program of research highlights the potential impacts that one's own and partners’ attachment and pre-sleep activity have on sleep.
A Comparative Genre Analysis Study of Scientific Articles Abstracts and AI-Generated Abstracts
(2024-06-18) Alqudah, A.R.; Zhi, Li; Spreng, Bettina; Makarova, Veronika
This study investigated and compared the rhetorical moves and steps employed in human-written abstracts of published articles with those generated by ChatGPT, 3.5 free version. The grammatical features and move patterns are also analyzed. The data for this study was collected by compiling corpus of 25 original research papers’ abstracts and 25 ChatGPT-generated abstracts. The ChatGPT abstracts were based on the titles of the 25 original abstracts collected prior, which were all published in Q1 journals from 12 different disciplines. Those disciplines are Psychology, Economics, Biology, Physics, Geology, Artificial Intelligence, Linguistics, Sociology, Agriculture, Mechanical Engineering, Management, and Sports Medicine. UAM CorpusTool (version 2.8), a corpus annotation tool, was used to annotate the texts based on a scheme developed by the researcher. Adapted from the IMRD (Introduction-Methods-Results-Discussion model for abstract writing, this scheme contains four moves, namely, Introduction-Methods-Results-Conclusion or IMRC, and a total of 21 steps. This adaptation reflected the rhetorical structures identified in the corpus. Four moves (Introduction, Methodology, Results, and Conclusion) were found to be employed in the abstracts written by human authors and those generated by ChatGPT. It was found that the Results move is the most frequent move in ChatGPT abstracts, while the Introduction move is the most frequent one in original abstracts. Significant differences were found in the frequency of the Introduction and Conclusion moves, and key literature findings step between the ChatGPT-generated abstracts and original abstracts. Such differences are likely due to the nature of the training data used for the AI model. As for the linguistic features, a high tendency was found in both ChatGPT and original abstracts to use past tense and simple present tense, content words, and active voice. However, significant differences found between ChatGPT and original abstracts indicated that simple present was employed much more in ChatGPT abstracts than in original abstracts, while the opposite was the case for simple past tense. Significant differences in the use of pronouns also showed that this part of speech is used significantly more in original abstracts. The move sequence patterns found showed that the patterns of ChatGPT is similar to that of typical research publications, while human abstracts display a variety of move sequence patterns. The findings of the study show that ChatGPT abstracts mimic original abstracts in terms of the rhetorical move pattern and in most of the steps used in each move. The researcher concludes the study with recommendations for future researchers based on the findings of the current study.
Multi-stage orogenic gold mineralization of the Pine Lake greenstone belt: evidence from the Seabee Gold Operation, Trans-Hudson orogen, northern Saskatchewan
(2024-06-17) Beaudin, Dallan; Partin, Camille; Ansdell, Kevin; Englington, Bruce; Lawley, Chris; Yuanming, Pan
The abstract of this item is unavailable due to an embargo.
Hunting for the Gathering: Terminal Pleistocene and Early Holocene Plant Resource Exploitation at the Quebrada Jaguay (QJ-280) Site, Southern Coastal Peru
(2024-06-17) Furlotte, Brett Alexander; Stuart, Glenn; Zarrillo, Sonia; Walker, Ernest; Swenson, Edward
Although plants were essential resources to human populations colonizing the desert Andean coast, paleoethnobotanical evidence associated with these events are scarce. While in some cases this is due to poor preservation, especially when considering the antiquity of the sites in question, the infrequent use of adequate recovery and analysis techniques is a bigger, preventable, concern. Ultimately, the lack of such evidence renders our understanding of the initial forays and subsequent settlement of the region incomplete.
In this thesis I report on novel macrobotanical evidence from the most recent excavations of Quebrada Jaguay (QJ-280), an archaeological site on the hyperarid coast of southern Peru with occupations spanning the Terminal Pleistocene to early Middle Holocene (ca. 14,000–8000 cal yr BP). Macrobotanical remains were recovered in situ during trowel excavation and from bulk sediment samples via a combination of fine meshed dry sieving and water flotation. Taxonomic identifications were carried out through stereoscopic observation and comparison of anatomical features with those of modern reference specimens. Comprised of woods, seeds, spines, leaves, fruit skins, and possible parenchymous tissues, the assemblage acquired includes a large total of 41 distinct macrobotanical types represented by a minimum of 37 taxa, the vast majority of which are sourced to lomas and riverine vegetation communities. These results show that a diverse range of wild plant resources (e.g., foods, medicines, fuels, craft and construction materials, animal fodder, and cultural items) from nearby coastal vegetation communities were available to and likely used by the sites inhabitants. In contrast to prior marine-centric subsistence characterizations, the wide range of plant foodstuffs recovered in particular, several of which were directly dated, suggests that the inhabitants enjoyed a broad spectrum rather than specialized diet. Finally, the presence of a few seasonal plant parts restricted in availability to particular times of the year indicate that edible plant resources, lomas cacti and riparian fruit tree seeds in particular, were likely harvested from winter–spring and spring–summer respectively. Based on this and other proxy evidence I suggest with caveats that Quebrada Jaguay, or the coast and interior lowlands at minimum, may have been occupied on a year-round basis.
Improving Deep Learning Classifiers for Plant Phenotyping using XAI Techniques
(2024-06-17) Mostafa, Sakib; Mondal, Debajyoti; Stavness, Ian; Sun, Shangpeng; Elshorbagy, Amin; McQuillan, Ian; Vassileva, Julita
The world's food security today is under threat and we need to improve the food production and nutritional values of the crops, invent climate-resilient foods, and plan better crop management to ensure food security. Precision agriculture is playing an integral role in food security by integrating technology to analyze large agricultural data to understand phenomic and genomic information, predict new breeds of crops, and improve crop production. Deep learning models are an important part of precision agriculture as they allow the analysis of large amounts of data and achieve impressive results. However, the deep learning models are of a black-box nature as they provide very little information about which features of the data contributed to the results and how such results were achieved. Explainable AI (XAI) allows researchers to investigate deep learning models and provide an explanation of the results which allows building trust in the model and helps improve and invent better models. In plant phenotyping, researchers are increasingly using deep learning models for analysis, but XAI techniques are being explored only recently. This thesis investigates the potential of XAI in plant phenotyping by leveraging XAI techniques for the selection and performance improvement of deep learning models with a focus on classification tasks in plant phenotyping.
I start by conducting a comprehensive review of the XAI techniques in plant phenotyping. The review provides a detailed overview of the state-of-the-art and traditional XAI techniques that are used in various domains of research. The XAI techniques are designed to improve the explainability of the deep learning models. Therefore, I reviewed the deep learning models that help plant scientists study plant phenomics. The use of deep learning models and XAI techniques in plant phenotyping is still in its early stages. I provide an overview of the application of deep learning and XAI in plant phenotyping and propose an XAI framework that may be used to ensure the use of XAI for plant scientists.
There are a few examples that use XAI techniques for the selection of a deep learning model in plant phenotyping. It is essential for a model to have the right complexity or depth relative to the dataset to achieve optimal performance. Therefore, I investigate ways to utilize XAI for selecting the model depth of deep learning classifiers performing plant phenotyping. I use a popular XAI technique known as Guided Backpropagation to visualize the learning of the intermediate layers of a deep learning model performing plant phenotyping tasks. I study the visualization of the feature maps to understand the relation between the capacity and the features learned by a model. I show that the shallow layers of a deep learning model learn more diverse features than the layers at higher depths. I designed a technique that leverages the Guided Backpropagation based visualization of the layers to provide insights into the model depth that should be used to achieve optimal performance for the dataset being used for model training.
Optimization of the deep learning models is considered to be an important part of building the model. I propose two novel approaches that utilize an XAI tool known as the activation pattern of the neurons to improve the accuracy of deep learning classifiers. In the first approach, I embed the neurons' activation probability in the training process as a loss function that increases the classification accuracy of state-of-the-art deep learning models on popular datasets by as high as 4.5%. In the second approach, I developed an auxiliary model based training process that achieves a classification accuracy improvement of as high as 5.2%. Unlike the loss function based approach, the auxiliary model based training is faster and can be used to boost the performance of a pre-trained model. I conduct extensive experiments with diverse deep learning classifiers on traditional and plant phenotyping datasets. The results show that the described techniques can potentially improve the accuracy of the classifiers in all cases.