21793 research outputs found
Sort by
Dialed Up Too High: The Aesthetics of Excessive Suffering in Hanya Yanagihara’s A Little Life
Get PDFThis thesis is interested in examining the rhetorical device of textual excess and how its use in Hanya Yanagihara’s 2015 novel A Little Life complicates plausibility, insofar as it concerns the genre of trauma fiction. I will specifically tend to the tension between the aesthetic objectives of A Little Life and how they conflict with the ethics of representation by defining “aesthetic” interests as those invested in the lavish glamours of the book and contrasting them with the aesthetic of suffering which constitutes Jude St. Francis’s being, and consequently bleeds into the rest of the novel. A Little Life is both renown and reviled for its famously distressed protagonist and his continuous misfortunes, such that it has joined the chorus of literature that favors pushing fiction past its textual and moral limits—not only was it published in a climate in which the preoccupation around authorial responsibility persists, but its events take this concern to a nearly unprecedented extreme. The result of this aesthetic marriage is both frustrating and urgent, in that Jude’s textual treatment subverts the reader’s expectations around believability as a spectator to his myriad abuses. Ultimately, even though the novel’s aesthetic choices trigger major morality points concerning the text’s commitment to describing and prescribing pain, the trigger itself is necessary in that it generates critical questions on representation and therefore what is represented is grounds for critiquing the nature and narrative of recovery from physical and psychological trauma
Integration of Inconsistency and Content Interaction with Deep Learning to Detect Fake Reviews
Get PDFIn this study, the challenge of detecting fake reviews in e-commerce is addressed through the application of natural language processing and deep learning techniques. The paper introduces two frameworks designed to identify fraudulent reviews, a critical concern due to their impact on consumer behavior and market dynamics. Central to this study is the exploitation of rating-sentiment inconsistency (RSI), a nuanced textual feature indicative of potential deception, aimed at enhancing the detection of fake content. Using a dataset of Amazon reviews, the paper evaluates two distinct approaches. The first method integrates RSI with word embeddings, specifically GloVe and Word2Vec, yielding an accuracy improvement of 3.07% for GloVe and 0.67% for Word2Vec, demonstrating the effectiveness of BiGRUs in capturing the sequential nature of textual data. The second method incorporates inconsistency features into Doc2Vec representations, achieving a 1.55% increase in accuracy compared to models without this feature. Both methodologies benefit from grid search optimization to fine-tune hyperparameters, enhancing model performance significantly. These combination methods not only underscore the importance of content-based feature integration but also demonstrate the practical application of inconsistency metrics in fake review detection. The observed improvements in model accuracy confirm the effectiveness of the proposed frameworks, providing new insights into enhancing online review system integrity and advancing natural language processing in commercial settings
Mum’s the Word: Leadership’s Role in Workplace Knowledge Hiding
Get PDFMum’s the Word: Leadership’s Role in Workplace Knowledge Hiding
Pouya Nikbakhsh
Knowledge is a critical strategic asset, yet deliberate knowledge hiding poses significant
challenges to collaboration and innovation. This thesis examines how supervisors’ rationalized
knowledge hiding influences employees’ tendencies to hide knowledge, using social learning
theory and perceived supervisor role modeling. Drawing on Leader–Member Exchange (LMX)
theory, it also explores whether high-quality supervisor–subordinate relationships moderate these
effects.
Data collected via Prolific from full-time employees across various industries were
analyzed using regression, mediation, and moderated mediation models. Findings show that
supervisors’ rationalized knowledge hiding strongly influences employees’ similar behaviors, but
not through explicit role modeling. Additionally, LMX quality did not significantly moderate
these relationships.
These results highlight the impact of leadership behaviors on workplace knowledge
dynamics. While supervisors’ direct influence is notable, role modeling and LMX alone do not
fully explain how knowledge-hiding norms spread. Organizations aiming to reduce knowledge
hiding should address supervisory behaviors and foster a culture of transparency and
collaboration
Environmentally friendly alternatives to MoS2-Lead based solid lubricants and anti-icing for aerospace applications: A combined experimental-computational study
No full textTo address two important challenges in aerospace industry which are 1) ice growth and 2)
friction and wear, two environmentally friendly materials such as hexagonal boron nitride (hBN)
and antifreeze protein (AFP) were pursued in this study. Molybdenum disulfide (MoS2)-based
solid lubricants commonly doped with lead (Pb) are widely used in industry. However, due to the
toxic nature of Pb, there is an interest in replacing it with more environmentally friendly material.
hBN is a potential alternative with a lamellar structure and a great lubricity property. Therefore,
MoS2-based solid lubricant containing five different concentrations of hBN (3.8, 5.7, 7.5, 9.5, and
11.5 wt.%) were applied on five 304L stainless steel substrates using spray bonding method to
explore their tribological performance. Reciprocating tribology tests were performed using a ball-
on-flat tribometer with alumina (Al2O3) counterface in atmospheric conditions. To characterize the
wear tracks, ex-situ analysis such as Scanning Electron Microscopy (SEM), Raman spectroscopy,
and Atomic Force Microscopy (AFM) were used. The MoS2-based solid lubricant with 5.7% hBN
revealed acceptable tribological performance. Additionally, to study the amount of ice growth on
a surface, atomistic molecular dynamics (MD) simulations were performed for three systems;
system A contains 3-layer hBN nanosheet, a layer of ice, and spruce budworm antifreeze protein
(sbwAFP), system B consist of 3-layer hBN nanosheet, an ice layer, and a non-antifreeze protein
found in Mycobacterium tuberculosis (MfpA), and system C includes 3-layer hBN nanosheet and
an ice layer without any biomolecules. The presence of sbwAFP and hBN presented less formation
of hexagonal ice in system A and C compared to system B
Detecting Concussion History in Athletes Using Pose Estimation and Machine Learning
Get PDFConcussions present a significant risk to athletes, with females exhibiting higher rates and prolonged recovery times than males. Current sideline concussion detection methods, such as the King-Devick test, suffer from validity issues, especially among young athletes, highlighting the need for more accurate and objective assessment tools. This study investigates the feasibility of using pose estimation technology, specifically Microsoft Kinect V2, to assess postural stability in varsity athletes with a concussion history. Inspired by previous research utilizing force plates, our study analyzes video recordings of athletes performing specific exercises to detect dynamic balance deficits. In a cross-sectional study of 444 varsity athletes in 2022 and 464 varsity athletes in 2023, results reveal significant differences in movement mechanics between concussed and control groups, with the Drop Vertical Jump (DVJ) exercise demonstrating the highest discriminatory power. Notably, concussed individuals exhibit longer time to stabilization (mean difference = 0.089 seconds, p = 0.046) during DVJ, indicating potential lingering balance impairments. While single leg squat and single leg hop exercises showed fewer discriminatory metrics than DVJ, they still provide valuable insights into balance capabilities. The DVJ was the most effective at distinguishing between injured and healthy male athletes, while the SLH was more effective for females and the SLS was equally ineffective for both males and females. Our research also studied 20 varsity athletes who sustained one or many concussions from 2022 to 2023 and compared their DVJ exercise metrics before and after injury. The results of this prospective study demonstrated few statistically significant differences between their 2022 and 2023 jumps for all computed metrics, suggesting that the occurrence of concussion(s) did not have a significant measurable impact on the athletes' jumping mechanics or dynamic balance for the DVJ over this period
Low-shot learning of substrate specificity on transmembrane transport proteins
Get PDFTransmembrane transport proteins are essential for cellular processes, selectively moving substrates across membranes. Traditional wet lab methods for detecting substrate
specificity, like binding and uptake assays, are costly and impractical for large-scale studies. Computational approaches, particularly machine learning (ML), could offer efficient
alternatives. State-of-the-art (SOTA) models to date can predict general groups of substrates carried by transporters, such as organic ions. The SOTA model TooT-SC achieves
a Matthew’s Correlation Coefficient (MCC) of 0.82 predicting 11 general substrate classes.
This research presents novel computational methods for predicting transported substrates using few-shot, one-shot, and zero-shot learning techniques to handle imbalanced datasets,
leveraging large language models (LLMs). These low-shot learning models enhance substrate specificity prediction.
An automatic pipeline is introduced to create machine learning (ML)-ready datasets for specific substrate groups, integrating the Chemical Entities of Biological Interest (ChEBI) and Gene Ontology (GO) databases to address the lack of annotated protein
sequence data. Initial studies confirm the effectiveness of transformer-based Protein Language Models (PLMs), adapted from natural language processing (NLP), in this context.
The research focuses on three key projects: TooT-Open-ICAT (Open-world classification of Inorganic Cations and Anions Transporters) predicts inorganic ion transport using
open-world classification; TooT-Triplet-SPEC (Triplet training for substrate SPECificity prediction) predicts specific substrates through metric learning; TooTranslator shifts from
classification to regression to predict substrates of uncharacterized proteins.
The TooTranslator model advances the SOTA by improving predictions for fine-grained classes accurately predicting 93 specific substrates with an MCC of 0.92. Furthermore, the models show promise in predicting true labels for unseen classe
The Myth of Religious Violence: Applying William Cavanaugh’s Theory to Jonestown, the Siege of Mount Carmel, and the Aum Affair
Get PDFThe 21st century has been witness to the proliferation of religious extremist groups and terrorist attacks committed by these groups. Much ink has therefore been spilled discussing the notion of religious violence, what it consists of, why it exists, and whether or not it should be understood as being particularly dangerous. In response to these inquiries, William T. Cavanaugh, a Catholic priest and theologian, has sought to deconstruct the notion of religious violence. Cavanaugh has argued that the term religious violence portrays acts of violence committed by religious groups as solely being the product of their religious beliefs, when such a thing is far from the truth, as no act of violence can take place in isolation. This thesis will prove Cavanaugh’s central argument correct through an analysis of 3 case studies that are typically understood as being significant acts of religious violence, the Jonestown massacre of 1978, the conflict at Waco between the FBI/BATF and the Branch Davidians in 1993, and the release of Sarin gas onto the Tokyo subway system in 1995 by the group Aum Shinrikyo. By analyzing these 3 case studies through Cavanaugh’s theoretical lens, this thesis will argue that the case studies in question should not be understood as incidents of religious violence, as to do so would be to ignore the fact that the violence undertaken was the product years of latent tension between secular authorities and the religious groups in question for reasons that had little to do with religious beliefs
Eco-Friendly Washing-Agent-Assisted Techniques for Removing Oil from Contaminated Shorelines
No full textOil spill accidents are significant environmental incidents that threaten environmental and ecological health. Among them, marine oil spills can cause severe damage to coastal residents and shoreline ecosystems. To enable effective response while minimizing additional environmental impacts, exploring and evaluating the potential use of low-toxicity chemicals for shoreline cleaning is crucial. This thesis begins with a comprehensive review of domestic and international literature, focusing on oil spill response techniques and the application of food-grade chemicals. Secondly, the study identified ovalbumin-based washing fluids as a promising option for shoreline oil removal. The effects of various environmental factors on the oil removal performance of ovalbumin washing fluids were examined. Different methods for the responsive separation of washing effluents were explored, and the separated precipitates demonstrated efficient decomposition using thermal and biodegradation methods. Thirdly, the oil removal performance of Accell Clean SWA under various low-temperature environmental conditions was assessed. Experimental results indicated that salinity and SWA concentration were the most significant factors influencing its effectiveness. Toxicity evaluations suggested that Accell Clean SWA is likely no more toxic than Corexit 9580 and PES-51, which have already been tested in real oil spill scenarios. The findings of this study provide valuable insights for developing potential green solutions for emergency responses to marine oil spills
Online Bipartite Matching under Markov Chain Model
Get PDFOnline bipartite matching (OBM) has a rich history in the literature of online algorithms, where it has been an influential problem inspiring many algorithms and techniques. This problem of obtaining a matching set of maximum size where vertices of one bi-partition arrive online has many real-world applications from kidney donor exchange to online advertising. This has led to the problem being studied under a variety of input models. In the adversarial model it is known that the tight competitive ratio is (among all randomized algorithms). Lower and upper bounds on competitive ratios better than are known for random order model, known and unknown IID models and other stochastic models, where a major open problem is to close these gaps.
One feature of the stochastic input models (e.g., known and unknown IID input models) under which OBM has been studied so far is the assumption of strong independence among the input items. One of the main conceptual contributions of this thesis is to introduce a stochastic input model that allows us to simulate limited dependence. In our model, input nodes are sampled from a Markov chain, and we refer to this as the Markov chain model. Introducing Markov chain significantly increases the complexity of analysis of algorithms by adding a number of parameters: initial distribution, transition probabilities, sampling size, etc, which leads us to concentrate on analyzing the problem for some specific families of bipartite graphs and Markov chains.
In particular, we study two algorithms \textsc{Non Adaptive} and \textsc{Adaptive Two Suggested Matching} under the Markov chain input model for parameterized versions of lazy random walks on -biregular type graphs. We give an alternative characterization of an offline optimal solution, , for these stochastic inputs, which allows us to calculate exactly (in the limit) the expected size of matching of . We then proceed to obtain tight bounds on the sizes of matchings obtained by the two algorithms under asymptotic conditions, which combined with the bound on , gives us tight competitive ratios of and for the \textsc{Non Adaptive} and \textsc{Adaptive} algorithms, respectively. These are competitive ratios for the classical lazy random walk Markov chain, where the probability of staying put is . We also derive exact formulas for competitive ratios with respect to lazy walks parameterized by -- the probability of staying put.
We believe these results, which use two disjoint matchings and regularity of graph degrees could be extended to type graphs of degree at most (for any constant ), and to bipartite graphs that admit several disjoint matchings of large sizes
Reinforcing Safety or Perpetuating Harm? Examining The Role and Effectiveness of Bill 151 and Sexual Violence Prevention Efforts in Montreal Universities
Get PDFThis thesis examines the effectiveness of Bill 151 and sexual violence prevention efforts
within Montreal universities. Bill 151, enacted by the Quebec government in 2017, mandates
higher education institutions to implement policies, training, and reporting mechanisms to
combat sexual violence. Through secondary data analysis, this research evaluates annual reports
and prevention policies from seven Montreal universities, assessing their adherence to Bill 151
and their impact on reducing sexual violence. The study reveals inconsistencies in policy
implementation, reporting quality, and prevention efforts among institutions. While some
universities demonstrate transparency and thoroughness, others exhibit gaps in data collection,
accountability, and survivor support. Findings indicate that institutional bias, lack of
standardized reporting, and low participation in training programs hinder the effectiveness of
these measures. Moreover, the persistence of underreporting and cultural barriers reflects
systemic issues within academic environments. By analyzing trends in reported cases and
prevention initiatives, the thesis underscores the need for intersectional, survivor-centered
approaches that address the root causes of sexual violence and institutional shortcomings.
Recommendations include enhancing training programs, improving data transparency, and
fostering a cultural shift to combat rape culture and promote accountability. This research
contributes to understanding how universities can better ensure campus safety and equity for all
students