21793 research outputs found
Sort by
Investigating Social Bias in LLM-Generated Code
No full textLarge language models (LLMs) have significantly advanced the field of automated code
generation. However, a notable research gap exists in evaluating social biases that may be
present in the code produced by LLMs. To solve this issue, we propose a novel fairness
framework, i.e., Solar, to assess and mitigate the social biases of LLM-generated code.
Specifically, Solar can automatically generate test cases for quantitatively uncovering
social biases of the auto-generated code by LLMs. To quantify the severity of social biases
in generated code, we develop a dataset that covers a diverse set of social problems. We
applied Solar and the crafted dataset to four state-of-the-art LLMs for code generation.
Our evaluation reveals severe bias in the LLM-generated code from all the subject LLMs.
Furthermore, we explore several prompting strategies for mitigating bias, including Chainof-
Thought (CoT) prompting, combining positive role-playing with CoT prompting, and
dialogue with Solar. Our experiments show that dialogue with Solar can effectively reduce
social bias in LLM-generated code by up to 90%.
Beyond single prompts, we studied social bias in multi-agent LLM workflows using
FlowGen, where agents act as requirement engineers, architects, developers, and testers.
The results show that the design of the workflow, the fairness-aware role instructions, and
the composition of the roles affect the fairness of the code.
Our findings demonstrate that social bias is a systemic issue in LLM-based code generation.
Solar offers a practical tool for assessing bias risks, tracing their origins, and applying
targeted mitigation strategies, including collaborative workflows
Efficient Synthesis of Symmetric Oligothiophenes via Decarboxylative and Suzuki Cross-Coupling Reactions
No full textOrganic semiconductors (OSC) have become an integral part of modern society and have found applications in various electronics such as organic photovoltaic cells (OPVC), organic light-emitting diodes (OLED) and flexible organic field-effect transistors (FOFET). Although poly-3-hexylthiophene (P3HT) is one of the most widely used OSCs, interest in oligothiophenes has surged recently as they can be used as effective models to study P3HT. The most widely used methods for the synthesis of oligothiophenes are Kumada, Stille and Suzuki cross-coupling (CC) reactions. These methods are well-explored and robust however, the Stille and Kumada processes come with a variety of disadvantages as well as environmental and health risks such as requiring harsh conditions and producing toxic metallic waste. In this work, we aimed to investigate and compare two different cross-coupling methodologies for the synthesis of oligothiophenes that do not produce harmful byproducts while also having simple and convenient reaction procedures. Using both decarboxylative cross-coupling (DCC) and Suzuki cross-coupling, symmetric oligothiophenes of lengths between 3 and 12 thiophene units have been sequentially synthesized at scales between 0.1 mmol and 1.4 mmol. Individual functionalized monomers were coupled to di-halogenated thiophene cores utilizing a double Pd-catalyzed cross-coupling reaction. The resulting oligomers were then di-brominated using multi-solvent systems with average reaction times of 30 minutes leading to complete conversions without the need for purification. Combining the DCC/Suzuki cross-coupling with the dibromination reaction, an odd sequence of 3, 5, 7, 9 and 11-unit oligothiophenes as well as an even sequence of 4, 6, 8, 10 and 12-unit oligothiophenes were sequentially synthesized in very good yields (70% to 92%). To our knowledge, 4 of the synthesized compounds have not been previously reported, namely 9T, 11T, DiBr-7T and DiBr-9T
Studying the Importin-regulation of Contractile Proteins in Cytokinesis
No full textCytokinesis is the physical separation of a dividing cell into two daughters. Multiple pathways regulate cytokinesis to spatiotemporally couple the segregation of chromosomes and the constriction of the contractile ring. RhoA, the master regulator of cytokinesis, is activated at the equatorial cortex by its activator Ect2. This interaction requires Ect2 interacting with the centralspindlin complex, although how the Ect2-centralspindlin complex is recruited to the membrane is unclear. After ingression, Ect2 signal at the midbody decreases, and Ect2 is presumably re-localized to the nucleus, but this has not been studied. It is also unclear whether decreasing Rhoa activity is required after ingression for abscission. In this work, we show that the polybasic cluster region (PBC) contains a nuclear localization signal (NLS) that binds to importins, that this NLS is required for cytokinesis. Mutating this NLS can abolish Ect2 recruitment at the membrane. We propose that importin-binding facilitates the recruitment of Ect2 at the cortex, potentially by favouring a conformation with higher affinity for lipids. We also show that the nuclear re-sequestration of Ect2 is required for the stability of the intercellular bridge and abscission. Mutating the central NLS, which mediates nuclear import, causes sustained RhoA activity at the intercellular bridge. We propose that nuclear sequestration may regulate other contractile proteins to promote midbody maturation and abscission. Finally, to identify other contractile proteins regulated by importins, we perform a TurboID assay and identify 10 potentially novel interactors of importin-b1 in mitosis, with potential roles in cytokinesis
Exploring Mother-Child Co-Regulation Across Interaction Contexts in At-Risk Mother-Child Dyads: Associations and Predictions to Child Externalizing Behavior Problems
No full textPsychosocial risk factors such as poverty and adversity negatively impact parent-child interactions, which are foundational for children’s socio-emotional development. Parenting stress and need for support, common in at-risk environments, also affect these interactions and have been linked to higher rates of externalizing behaviours in children. This study examined co-regulation, a dynamic process through which parent and child mutually influence each other’s emotions, cognitions, and behaviours, in psychosocially at-risk mother-child dyads. Participants were 111 mother-child dyads from the Concordia Longitudinal Research Project. Dyads were observed during three tasks: a puzzle task, a compliance-based command task, and an interference task simulating maternal unavailability. Five types of co-regulation from the Revised Relational Coding System were coded: symmetrical, unilateral, unengaged, asymmetrical, and disruptive. It was found that dyads engaged in more symmetrical co-regulation during the puzzle and command tasks, and more unengaged and unilateral co-regulation during the interference task. Moreover, more time spent in symmetrical co-regulation during the puzzle task was associated with greater externalizing behaviours in children. In contrast, more unilateral co-regulation was associated with fewer externalizing behaviours in middle childhood. Maternal parenting stress was also a predictor of behavioural problems. Findings highlight the context-dependent nature of co-regulation; dyads engaged in different types of co-regulation depending on the task. Further, in at-risk dyads, symmetrical co-regulation may reflect mutual dysregulation, rather than adaptive coordination, while disengagement may be protective. Results have implications for promoting adaptive dyadic regulation in at-risk families through parenting interventions and prevention programs
Exploring Disability Identity Through Reflexive Songwriting: A Heuristic Self-Inquiry
No full textMusic therapists have long supported autistic and neurodivergent (ND) individuals, yet support needs are too often defined through deficit-based models rooted in observable behaviours described in the Diagnostic and Statistical Manual of Mental Disorders (DSM). Recently, disabled, autistic, and ND music therapists have begun challenging this framing, calling for approaches that reflect the diversity and lived realities of those we serve. Disability experience itself offers crucial insight into how music therapy might better honour neurodivergent ways of being and foster more inclusive care.
This heuristic self-inquiry aimed to contribute to these first-voice perspectives by exploring what insights might emerge when the researcher, a neurodivergent Certified Music Therapist, engaged in reflexive songwriting to claim a disability identity. The study also sought to examine how this process could deepen her understanding of disability and neurodivergence as identity, inform her ability to identify and articulate support needs, and shape her clinical practice.
Data were collected through journaling and reflexive songwriting, guided by Moustakas’ six phases of heuristic self-inquiry, and analyzed using qualitative content analysis. Three main categories emerged: grappling with expectations, resisting expectations, and the courage to look ahead—each with related subcategories. A creative synthesis took the form of a poem that spoke directly to normativity, imagined as a person or collective entity, through which key findings and a forward-looking vision were articulated.
The thesis concludes with a discussion of implications for practice and the researcher’s reflections on the multilayered value of reflexive songwriting in exploring disability identity
Self-Constitution in Foucault and Schürmann: The Question of a Normative Ground for a Post-Metaphysical Ethics
No full textFoucault’s account of self-constitution in his late work offers strategic insights into how a subject can constitute themselves as a practical or transgressive subject but, fails to show how that subject can be constituted ethically. Foucault’s methodological commitment to reformulating transcendental conditions as historical conditions means that he cannot show how the subject can be constituted according to any binding normativity that could offer direction or orientation for that self-constitution. In his attempt to integrate Foucault and Heidegger, largely neglected in the secondary literature, Reiner Schürmann attributes this to Foucault’s failure to ‘step back’ to the transcendental level of Heidegger’s history of being, but does not develop this criticism any further. I argue that Schürmann’s interpretation and elaboration of Heidegger’s insights resonates with Foucault’s critical work in that it offers a non-foundational, ‘anarchic’ ontology that can provide a quasi-normative ground in the form of critical responsibilities, and that Schürmann’s account of the ultimate double bind of natality and mortality situates and contextualizes self-constitution in a way that corrects the overly voluntaristic emphasis on the freedom of self-transformation that has been widely criticized in Foucault’s work
Detection of Vegetation Proximity to Power Lines: Critical Review and Research Roadmap
No full textThe resilience of power distribution systems is crucial for maintaining the stability and functionality of modern societies. The proximity of natural vegetation to power lines poses significant risks, particularly when combined with adverse weather events. This review paper examines state-of-the-art methods for detecting and managing tree proximity to power distribution lines using advanced machine learning (ML) techniques, including deep learning (DL) applied to remote sensing data. The complex interactions between adverse weather conditions and power outages caused by tree encroachment are explored. The potential of AI-driven monitoring systems to enhance vegetation management strategies, thereby mitigating the risks associated with tree-related power outages, is underlined. A significant gap in the literature is identified, with few studies specifically addressing the application of ML/DL for dynamic monitoring of tree proximity to power lines. A detailed comparative analysis of existing methodologies is provided, emphasizing the unique contributions and limitations of current approaches. Future research directions, including the development of more sophisticated ML/DL models and the integration of multi-sensor data, are outlined. This review serves as a critical resource for researchers, utility managers, and policymakers aiming to improve the resilience and reliability of power infrastructure management
Honouring The In-Between: Video Collage-Based Art Therapy Intervention for Second-Generation Canadian Youth
No full textThis philosophical inquiry explores the therapeutic potential of video collage as an art therapy intervention for second-generation Canadian adolescents navigating identity development and integration. Adolescence is widely recognized as a key period of identity development, with second-generation bi- and multicultural youth often encountering additional complexity as they negotiate multiple cultural selves: identities associated with both heritage and dominant cultures (Giguère, Lalonde, & Lou, 2010). As many adolescents engage with digital media as an everyday means of communication and self-expression (Varadi, 2025), video-based intervention offers a familiar and adaptable creative medium for exploring identity. This research positions video collage as a tool uniquely suited to accommodate layered, fluid, and non-linear expressions of self, allowing youth to represent and affirm their identities in ways that hold complexity without requiring reduction or closure. In dialogue with literature on art therapy, therapeutic filmmaking, and Participatory Action Research approaches, this study emphasizes the importance of youth agency, power-sharing, and culturally responsive practice when implementing video-based interventions. Key considerations for art therapists include ethical handling of digital materials, attention to power dynamics, and ongoing reflexivity regarding cultural and systemic influences shaping identity. Findings suggest that video collage holds significant promise in supporting self-authorship, emotional expression, and pride in multifaceted identities among second-generation youth, while also demonstrating a need for further empirical research to develop structured frameworks for clinical practice
The Annihilation of Time and Space
No full textThe Annihilation of Time and Space is an interdisciplinary project exploring memory, queer identity, vulnerability, and the power of the gaze through photography, film, light, and installation. Rooted in a formative childhood experience of pleasure, exposure, and shame, the work revisits the suburban backyard as a site of intimate reckoning. Through video projections, light boxes, composite photographs, and a reappropriated family photo album, the project interrogates the instability of memory, the relational dynamics of looking, and the construction of masculinity and the objectified self. Viewers oscillate between roles of observer and observed, experiencing intimacy, vulnerability, and the fragility of identity. The project transforms early experiences of shame into moments of recognition, care, and possibility, collapsing temporal and spatial distance between past and present selves. Ultimately, it envisions looking—not as surveillance, but as an act of attention, tenderness, and love
Examining the Importance of AI-Based Criteria in the Development of the Digital Economy: A Multi-Criteria Decision-Making Approach
Get PDFAs one of the main pillars of global transformation in the contemporary world, the digital economy helps create new economic and business opportunities through new technologies. In addition to improving efficiency and reducing costs, this transformation plays a vital role in the economic growth and development of various countries. Artificial intelligence, as one of the key technologies in the development of the digital economy, has a profound impact on optimizing processes, increasing productivity, and enhancing customer experience. By processing big data and providing advanced analytics, this technology makes economic decisions faster and more accurately and affects various sectors of the digital economy. In this regard, 20 key AI-based criteria in the development of the digital economy were extracted from a review of previous studies and were placed in four general categories. The four general categories include structural, organizational, technological and economic. Hesitant Fuzzy Best Worst Method (HF-BWM) was used to rank the AI-based criteria in the development of the digital economy. “Investing in innovation (C16)”, “Potent processing capabilities (C1)”, “Process automation and intelligence (C11)”, “Identifying growth opportunities (C6)” and “Adapting business models to changes (C7)” ranked one to five, respectively. Managers in the digital economy should pay attention to investing in innovation and strengthening processing infrastructure to exploit new technologies and make more accurate decisions. Process intelligence, identifying new areas of growth and adapting the business model to market changes also help improve efficiency, reduce costs, exploit new opportunities and make organizations stable in the face of rapid changes and increasing competition