12466 research outputs found
Sort by
Immigrant Family Cultural Conceptualizations of Disability and a Pathway for Special Education Teacher Preparation
No full textThe United States is a country known for considerable cultural, linguistic, and racial diversity. These contexts are present within public-school settings, which are comprised of students and families from diverse backgrounds holding different experiences and perceptions of education. The diversity present within public-school classroom settings has important implications for the field of special education, especially in regard to educator and family partnerships, a foundational aspect of special education. In this qualitative dissertation, I explore (a) empirical research on the topic of immigrant family cultural perceptions of the term disability; (b) cultural conceptualizations of disability and school experiences of Korean immigrant legal caregivers of students with disabilities; and (c) a possible pathway to support special education teacher preparation in regards to cultural and global contexts in special education. Findings reveal various cultural beliefs regarding disability and the impact of an immersive experience studying an international education system. Implications for research, teacher preparation, and policy are discussed
Ground-based Light Curve Follow-up Validation observations of TESS object of interest TOI 5886.01
Full text link“NASA's Transiting Exoplanet Survey Satellite (TESS) is using a transit photometry method to look for dips in the brightness of the nearest 200,000 stars to scan for possible exoplanets. This paper presents the results of follow-up ground-based observations of TESS Object of Interest (TOI) 5886.01, also identified as TIC (TESS Input Catalog) 15682927, and aims to determine if the transit detected on TOI 5886.01 represents a potential exoplanet. Utilizing Python and Jupyter Notebook, data from TESS, and analyze our own data from GMU Observatory (collected on 06/18/2024) with AstroImageJ, our evaluation suggests that TOI 5886.01 is likely a hot Jupiter.
Extracting Clinically Relevant Information from Text Using Pre-trained Language Models
Full text linkText containing medically relevant information is available in various forms across different platforms. Electronic Health Records (EHRs) describe medical conditions, treatment, and outcomes. Text on social media platforms like Reddit contains posts where users describe their experiences with various medical conditions and ask questions or make claims about the efficacy of treatments on their health. There is a clear need to leverage EHR data for secondary uses, including precision medicine, cohort selection for experimental treatments, and public health. On the other hand, systems that utilize social media texts can help us better understand patient experiences.EHRs contain patient information from structured and unstructured data intended to document patient care. Certain types of patient information are better characterized in unstructured data, including medication changes, their rationale, and social determinants of health (SDoH). These patient data are often represented in narrative form rather than structured data since free text allows for detailed descriptions. In contrast, user text on social media platforms can be intended to exchange information or discussion and are solely in narrative form. Natural language processing (NLP) and Information Extraction (IE) methods can be applied to extract relevant information from EHRs and social media user text for specific downstream use cases. Data-driven IE models utilize annotated data for training and evaluation. Multiple factors affect the use of IE models, including the high cost of human annotation and the lack of open-source annotated corpora for intended downstream use cases. Even with relevant annotated data available, the performance of IE models is limited by the complexity and ambiguity of language. Narrative EHR text can be technical and complex, written for experts in medicine. Medically relevant information represented in social media text can have complex language structures, including words that occur relatively infrequently in scientific text. These challenges are magnified when phenomena of interest are sparse and granular. This work explores IE for EHRs and social media text. We study EHR-related IE through medication change and SDoH-related information. The medication change IE work focuses on building models to resolve ambiguity in multiple medication change descriptions. We utilize an existing annotated corpus, Contextual Medication Event Dataset (CMED), and build novel IE models that can reliably extract medication change information with high performance. The proposed novel architectures show state-of-the-art performance on medication change extraction, including challenging descriptions that contain co-occurring medications with different change attributes. The exploration of SDoH IE focuses on addressing challenges associated with the limited availability of annotated corpora and reducing human annotation costs through novel IE methods that require significantly less training data. As part of SDoH exploration, we utilize two annotated corpora, Social History Annotation Corpus (SHAC) and Pediatric Social History Annotated Corpus (PedSHAC). We present state-of-the-art SDoH IE models that extract the annotated phenomena with high performance and require significantly less training data. We show that sequence-to-sequence architectures performed significantly better than the prior work when extracting pediatric SDoH. Our IE approaches could extract SDoH with limited training data, achieving performance on par with supervised learning models. The exploration of IE for social media text focuses on extracting medical condition-related information, including patient experiences and claims, along with granular information on patient population, interventions, and outcomes (PIO). We study these phenomena on an existing annotated corpus, Reddit Health Online Talk (RedHot). We build high-performing IE models using domain-specific and mixed-domain models that utilize external knowledge to extract the annotated phenomena. We show that domain-specific models perform better in identifying medical condition-related entities, while general-domain models are as good as domain-specific models in understanding patient experience-related descriptions. Including external knowledge, specifically disease information, helps improve the detection of patient population types among the users, and including knowledge about chemical names helps to better identify entities related to clinical interventions
2024 Assessment of Virginia’s Information Ecology of the Disability Services System
Full text linkAccess to disability services information depends on many factors, from an individual’s digital literacy, social connections and physical mobility to the interface design of websites. However, it is also true that the availability of disability services information (e.g., how to apply for a Medicaid Waiver) and how such information is managed and provided to end users in Virginia are also critical factors that shape people’s information access. This assessment focuses on understanding the latter, namely, the “information ecology” of disability services in Virginia. Specifically, we examine how service providers, both governmental and non-governmental agencies, manage information about disability services, the challenges staff face in providing information, and how different types of information are communicated and shared across different providers.Funding for this project/product was supported, in part, by the Virginia Board for People with Disabilities, under grant number 2301VASCDD, from the U.S. Administration for Community Living (ACL), Department of Health and Human Services, Washington, D.C. 20201
Searching for Identity, Belonging, and Community: Second-Generation Muslim American Participation in Muslim Spaces
Full text linkSecond-generation Muslim Americans continue to grapple with questions of belonging and identity as they encounter discrimination, marginalization, Islamophobia, and otherization in post-9/11 America. In the wake of the resurgence of right-wing politics, which has given rise to anti-immigrant and anti-Muslim sentiment in American society, young Muslims are compelled to navigate and negotiate the meaning of their racial, ethnic, and, especially, religious identities. Drawing on in-depth interviews and field observations, this dissertation examines the sense of belonging, identity, and integration of young second-generation Muslim Americans in the Washington, D.C., Maryland, and Virginia area. The study investigates the construction of identity and belonging through participation in Muslim spaces, including mosques, Muslim student associations, and faith-based organizations. The study finds that unlike the suburban, monoethnic mosques of their immigrant parents, which cater to the religious and social needs of first-generation Muslim immigrants, young Muslim Americans turn to more inclusive Muslim spaces to build community, friendships, and belonging. Among them, faith-based organizations, which are both religious and social in nature, have emerged to fill the gap left by mosques and Muslim student associations. In these spaces, young professional Muslim Americans interact with Muslims of diverse racial, ethnic, and socioeconomic backgrounds, developing friendships and a sense of community and belonging. Moreover, these spaces allow the broadening of their interest in their faith through discussions, debates, and reinterpretations of Islamic teachings that may have been contentious in traditional settings such as mosques. Hence, faith-based organizations provide both physical and symbolic spaces of belonging, resulting in a pan-Muslim American identity that bridges racial and ethnic divisions. Ultimately, identity formation and a sense of community within these spaces lead to the construction of an 'American Islam' that aligns with American values of personal freedom and social justice
Discordance Over Health, Race, and Politics in American Protestant Churches: A Case Study of Fairfax Church’s Use of Conflict Prevention & Resolution to Promote Unity Over Uniformity
Full text linkThis study examines polarization affecting American Protestant churches, particularly the alignment of extreme right-wing partisanship with Christianity. By integrating perspectives from religious studies and conflict resolution, it conducts a case study of Fairfax Church, a multi-ethnic megachurch located twenty miles from Washington, D.C. The research assesses the church's leadership in employing conflict prevention and resolution tools during culturally and politically divisive events, including the COVID-19 pandemic, the death of George Floyd, and the January 6th insurrection. Drawing on qualitative interviews with church leaders and quantitative survey data from the congregation, the findings demonstrate that Fairfax Church is effectively contributing to a less polarized society by prioritizing unity and peace while also pointing out the inadequacies where there is room to grow. The study offers actionable recommendations for church leaders and individual Christians, such as prioritizing the Gospel through scripture, resisting the influence of political culture, acknowledging divisive events, embracing an "exile" mentality, focusing on ethical means rather than ends, understanding scripture, humanizing opponents, rejecting fear, and pursuing unity. Despite its limitations, this research contributes to the growing body of literature on religious peacebuilding by highlighting the role of religious principles in addressing political and cultural polarization in America
The Deep Hypocrisy: How U.S. Presidents Enacted Violence on the “Dangerous” Latin American Immigrant
Full text linkThe purpose of this historical inquiry is to reveal the way presidents have used their speeches as a tool of constructing the narrative of the dangerous Latin American immigrant while their administrations actually inflicted violence on Latin American immigrants and Latin American countries. Narratives will be analyzed through three case studies focused on President Dwight D. Eisenhower, President Ronald Reagan, and President Trump. In order to analyze the way these presidents characterized Latin American immigrants in comparison with their immigration and foreign policies, this study utilized qualitative case study methodology. These case studies will cover topics of immigration, racism, and Latin America
The Impact of David Baker's Pedagogy in the Field of Jazz Improvisation Education
Full text linkABSTRACT THE IMPACT OF DAVID BAKER’S PEDAGOGY IN THE FIELD OF JAZZ IMPROVISATION EDUCATION Michael M. Walker, Sr., Ph.D.George Mason University, 2024 Dissertation Director: Dr. Charles R. Ciorba The Purpose of this study was to address how and to what extent David Baker’s jazz pedagogy impacted the field of jazz improvisation education. The study examines interviews of six leaders in university jazz programs and/or professional performing musicians with firsthand knowledge of Baker’s life and teachings. All are established pedagogues and performers in the field of jazz. Profiles of the interview transcripts were created for this qualitative study to alleviate the idiosyncrasies of speech in the interviews. Thematic connections were derived with the use of headings from the interview questions. The interview data in the profiles exposed empirical information that was used for interpretive analysis. Results indicated Baker’s educational methods and books had a substantial impact in the field of jazz education. Findings also revealed that his relationships with his students had a significant impact on jazz education. Evidence from the participants interviewed uncovered a connection that transcended his methods and pedagogical ideas. It was concluded that David Baker left a lasting heritage for jazz education. His seminal methods and pedagogy provide a living legacy. Keywords: pedagogy, Bebop, lingua franca, transcriptions, definitive recording, jazz improvisatio
Understanding User Privacy in Immersive Multimedia Platforms
Full text linkVirtual reality (VR) technology has become increasingly popular in recent years. It is used in various applications, including entertainment, healthcare, and military sectors, where sensitive data is handled. However, as a system at its infancy, VR platform's protection of the sensitive data is unknown. Therefore it is vital to identify the potential vulnerabilities in the system, and devise robust mechanisms to protect the data. We focus this study on the human computer interaction (HCI) component in VR, as this is the most different part compared with traditional devices such as smartphones and desktops.In VR, users interact with virtual content through a head-mounted display (HMD) and hand controllers, often exhibiting intensive eye movements to view the panoramic virtual content. Unlike traditional devices, VR system track user movements through various sensors embedded in the HMD and controllers, while completely blocking users' view of the physical world. These differences in VR HCI raise questions about efficiency of existing privacy protection methods, vulnerabilities stemming from the user’s blocked view of the physical world, and the feasibility of utilizing eye movement patterns for user identification in VR. In response to these questions, our projects include developing a privacy protection method that utilizes human visual system (HVS) features, exposing a keystroke inference attack in VR that eavesdrops controller acoustic emanation, and achieving seamless user identification in VR utilizing eye movement data captured by the HMD’s eye tracker. In the first work, we explore the HVS for reliable and unobservable VR HMD authentication. Our system, OcuLock, explores HVS biostructure and unique HVS features triggered by immersive VR content to enhance authentication stability. OcuLock is empowered by an electrooculography (EOG) based HVS sensing framework and a record-comparison driven authentication scheme. Experiments through 70 subjects show that OcuLock is resistant against common types of attacks such as impersonation attack and statistical attack with Equal Error Rates as low as 3.55% and 4.97% respectively. More importantly, OcuLock maintains a stable performance over a 2-month period and is preferred by users when compared to other potential approaches. In the second work, we expose a placement-flexible keystroke inference attack in VR by recording and analyzing the clicking sounds of the moving hand controller during keystrokes. As the first acoustic inference attack in VR, our system, Heimdall, overcomes unique challenges unaddressed by previous acoustic attacks on physical keyboards and touchscreens. These challenges include differentiating sounds in a 3D space, adaptive mapping between keystroke sound and key in varying recording placement, and handling occasional hand rotations. Experiments with 30 participants show that Heimdall achieves key inference accuracy of 96.51% and top-5 accuracy of 85.14%-91.22% for inferring passwords with 4-8 characters. Heimdall is also robust under various practical impacts such as smartphone-user placement, attack environments, hardware models, and victim conditions. In the third work, we propose a seamless user identification method based on eye tracking data acquired during free viewing. Our system, OcuTrack, utilizes user-specific patterns of the interplay between eye movements and visual content to identify users when they view different types of content. OcuTrack is empowered by an eye movement-visual content decoupling scheme and a comparison-driven user identification framework. The effectiveness of OcuTrack is demonstrated through extensive evaluations where our model achieves an impressive F-score of 0.9166 with a minimal fluctuation within 0.04 when presented with different visual content. OcuTrack is also robust under various practical impacts such as different hardware models and user conditions
An Informed Exploration of the Technology-Supported Learning Needs and Preferences of First-Time Fathers in the Perinatal Period
Full text linkThis study employed qualitative research methods to conduct an in-depth learner needs analysis of the learning motivations, needs, and preferences of first-time fathers in the perinatal period. Occurring within the informed exploration phase of a Design-Based Research approach, the study aimed to understand the unique learning experiences of first-time fathers to inform recommendations for the instructional design and development of a resource to support fathers in their transition to fatherhood. Guided by Knowles’ adult learning theory, this study examined fathers’ learning goals, motivations, and preferences for technology-supported learning in the perinatal period. A qualitative content analysis of an online community for new fathers informed the development of semi-structured interview questions. Data were collected through interviews with 19 fathers who were either expecting a child or having a child born within the past two years. Interview transcripts were analyzed using focused and pattern coding to identify emerging themes related to fathers’ motivations and goals for learning, their learning focus and priorities, father-to-father connections and support, their use of technology for learning, father-specific learning content, and fathers’ learning preferences. Personas were developed based on findings from the interviews to inform design recommendations. This study shows that fathers have distinct learning needs, preferences, and experiences, particularly in using technology to support their learning in the perinatal period. Fathers are intrinsically motivated to learn and prioritize learning practical skills, particularly when lacking prior knowledge and experience. Their learning preferences and activities changed over time, shifting from proactive to just-in-time learning. Fathers utilized technology in various ways to support their learning, including accessing social media for short, digestible pieces of microlearning that could be incorporated into their daily lives. Additionally, fathers expressed an interest in father-specific content but found the stereotypes and low expectations presented in much of the existing content offensive and unhelpful. The most significant finding of the present study is the crucial role of father-to-father connection and support, a preference that impacted fathers’ learning goals and preferences across the perinatal period