Scholarly Works, University Libraries

Permanent URI for this collection

https://hdl.handle.net/10919/25336

Research articles, presentations, and other scholarship

Browse

Now showing 1 - 20 of 583

Community-Owned-and-Operated: Amplifying Cultural Heritage through Inter-Institutional Collaboration
Kinnaman, Alex; Palazzo, Ashley (2024-11-15)
The Greater Southwest Virginia Digital Collective (GSDC) is a community-owned collective steered by a volunteer community advisory board composed of community members from the region and Virginia Tech University Library (VTUL) faculty members that reviews, approves, and champions community collections to be added to the Virginia Tech Digital Library. The challenge GSDC aims to address is the gap between well-resourced institutions and smaller cultural heritage organizations by providing community-tailored assistance in processing and describing collections, digitization and consultation, and depositing material into an access and preservation repository. This roundtable will consist of two GSDC members representing both a community organization and VTUL to discuss the relationship-building process, successes and challenges, and sharing the community-driven model of GSDC.
Dancing with Donors: Trust-Building Across Gaps of Curation Priorities
Munshower, Alan; Kinnaman, Alex (PubPub, 2024-09-16)
Virginia Tech University Libraries (VTUL) serves a range of cultural heritage, academic, and local communities aligned in the goal to “get stuff online and accessible.” Despite the same overarching goal, the specific requirements from each party to reach that goal do not always overlap. The initial dance of negotiation between library and donor collaborations sets the tone for the ongoing relationship between the two. Across the departments in VTUL that manage such relationships with donors and curate digital collections, there are common trends emerging in barriers and observations with building relationships, and also with the concessions, compromises, and adjustments made to meet the curation needs of both parties. There are noted gaps in priorities and knowledge of curation processes, expectations around the understanding of digital collections, communication and roles and responsibilities, and resource understandability and availability. This paper specifically addresses relationships with donors and that impact on the subsequent work resulting from agreement with both parties. Continuing the iPRES conversation around community archiving and successful collaborations, the authors of this paper look critically at their partnerships with donors of digital material. This paper aligns with the conference theme “Start 2 preserve” in that it both addresses the barriers to entering the digital preservation landscape for the non-librarian community, and the barriers of digital preservation practitioners in aligning collaborator needs with digital curation needs. The authors focus on spotlighting the learning curve present on both sides of the work of community archiving. In recognizing recurrent gaps in understanding, this paper aims to be a part of a larger conversation on how community partnerships can blossom with built trust and understanding, coupled with robust planning and technical capability.
Creation of an Incentivized Course for Managing Your Online Scholarly Identity
Miles, Rachel A.; Mazure, Emily S. (2024-11)
Librarians at a large research-intensive university in southwest Virginia in the United States developed an online asynchronous course on how researchers can manage their online scholarly identity. It explains the importance of understanding and efficiently using scholarly identifiers and profile systems and also guides participants through the process of creating and maintaining scholarly profiles and identifiers, with a goal to have participants complete specific activities. To encourage completion of those activities the course was designed with specific incentives; for example, credits earned from the course can be used by faculty to complete the university’s computer-refresh program, which enables them to acquire a new computer after a four-year period. The content was developed in the institutional professional development Canvas platform and was thus available internally to faculty and graduate students. Participants can self-select which modules to complete. Additionally, participants can submit proof and receive credit for completing specific tasks like registering for ORCID, linking IDs across profile systems (e.g., Scopus, Google Scholar Profile, Elements profile, etc.), completing profile details like education, academic positions, scholarly works, and so on. This course is intended as a pilot that we expect to expand upon. The future goals of the course will be to cover two additional strategies for boosting online scholarly visibility: increasing discoverability and openness of scholarship and promoting work through social media and other online channels.
Prompting Best Practices: How Are Libraries or Their Home Institutions Creating, Sharing, Applying, and Adapting GenAI Policies?
Pannabecker, Virginia (2024-06-11)
Join me for this 45-minute discussion-based review of institutional policies created by libraries or their larger institutions on the use of Gen AI in teaching, learning, and research. During the first 15 minutes, I will share a selection of policies from 5-10 institutions, highlighting examples of commonalities and key differences, including how each policy addresses ethical aspects of using AI in the institution's context. The second 15 minutes will be breakout small group discussions of the example policies, policies participants are aware of or use at their own institutions or at institutions they’re curious about; and each group will have an online space to jot down notes and add links to policies or resources they discuss. The last 15 minutes will include a 1-2 minute report back from each group about useful aspects they found in the example policies or other policies discussed in their group, questions or concerns about policies discussed, examples of applying such policies at their institutions, examples of how to stay up to date with changes in Gen AI usage and make nimble adjustments to policies, or recommendations and comments for Gen AI policies going forward. We’ll conclude with a wrap up and links to the shared discussion documents for reference.
Contributions of OER toward Student Success (VT 2024)
Walz, Anita R. (2024-11-05)
This presentation provides an overview regarding the value of open educational resources toward student success and university priorities. Presented at the University Libraries at Virginia Tech Library Forum on November 5, 2024.
Extension Microfilm Digitization Project: Putting History Into Our Hands
Haugen, Inga; Westblade, Julia; Russell, Meagan (2024-05-07)
The Virginia Cooperative Extension microfilm digitization project aims to create digital copies of and provide access to the agricultural reports of the state of Virginia. These primary source reports consist of the work of extension agents at the county-level from 1908 to 1968 for men and women from white communities and communities of color, including information regarding production and salaries. This paper will discuss the process of digitizing 141 reels of microfilm and making the contents accessible to researchers. The paper will highlight the methodologies and challenges experienced during the process as well as the importance of the data uncovered in the documents. It will give an overview of the effort it takes to provide access to primary resources that researchers need to uncover untold stories. Digitization of the Microfilm The original documents were scanned onto microfilm in the 1960s. The digitization lab at Virginia Tech's Newman Library has digitized, reformatted, sorted, and combined into text-searchable PDFs over 100,000 pages of county-level reports adhering to FADGI standards. The team had to document progress as the project moved through several stages of production before members of the team sorted through these PDFs to create item-level metadata to ensure the reports are findable and searchable. Document Overview/ Importance This set of microfilm was the most complete set in the state and in WorldCat, and had a reel guide of the counties and years for only 86 of the 141 reels. This project will bring to light individual reports, the authors, and the extension work that was happening in the whole state from 1908-1968. Because the authors include women and Black extension agents, this work brings local history into the hands of the communities we currently serve. As an example, a technician saw a report about her partner’s grandfather while processing the collection.
Enhancing Digital Twins with Human Movement Data: A Comparative Study of Lidar-Based Tracking Methods
Karki, Shashank; Pingel, Thomas J.; Baird, Timothy D.; Flack, Addison; Ogle, J. Todd (MDPI, 2024-09-18)
Digitals twins, used to represent dynamic environments, require accurate tracking of human movement to enhance their real-world application. This paper contributes to the field by systematically evaluating and comparing pre-existing tracking methods to identify strengths, weaknesses and practical applications within digital twin frameworks. The purpose of this study is to assess the efficacy of existing human movement tracking techniques for digital twins in real world environments, with the goal of improving spatial analysis and interaction within these virtual modes. We compare three approaches using indoor-mounted lidar sensors: (1) a frame-by-frame method deep learning model with convolutional neural networks (CNNs), (2) custom algorithms developed using OpenCV, and (3) the off-the-shelf lidar perception software package Percept version 1.6.3. Of these, the deep learning method performed best (F1 = 0.88), followed by Percept (F1 = 0.61), and finally the custom algorithms using OpenCV (F1 = 0.58). Each method had particular strengths and weaknesses, with OpenCV-based approaches that use frame comparison vulnerable to signal instability that is manifested as “flickering” in the dataset. Subsequent analysis of the spatial distribution of error revealed that both the custom algorithms and Percept took longer to acquire an identification, resulting in increased error near doorways. Percept software excelled in scenarios involving stationary individuals. These findings highlight the importance of selecting appropriate tracking methods for specific use. Future work will focus on model optimization, alternative data logging techniques, and innovative approaches to mitigate computational challenges, paving the way for more sophisticated and accessible spatial analysis tools. Integrating complementary sensor types and strategies, such as radar, audio levels, indoor positioning systems (IPSs), and wi-fi data, could further improve detection accuracy and validation while maintaining privacy.
Public Domain and Paywalled: Journal Articles Authored or Co-authored by U.S. Government Employees
Young, Philip; Ghaphery, Jimmy (University of Kansas Libraries, 2024-08-21)
Academic journal articles authored by U.S. government employees are assumed to be in the public domain, though journals vary in communicating this status, and access is often not provided. To document this situation, between September 2020 and March 2021 we collected and analyzed copyright statements from a random sample of articles in PDF published in 2019 by authors affiliated with two U.S. government agencies. 13% of the sampled articles had a copyright statement indicating the U.S. public domain or U.S. government authorship. 42% of the published versions of the sampled articles were behind a paywall. Even when all authors of an article were U.S. government employees, 29% were labeled in the U.S. public domain, and 66% were behind a paywall. While copyright notices are not required, notice provides legal certainty on the usage of journal articles, which are shared among scholars, added to bibliographic managers, and posted to websites and repositories. Journal articles authored by U.S. government employees may be a source of open access that has not been fully realized, and uniquely, a retrospective source of access for scholarship. We suggest best practices for journal publishers, as well as possible actions by U.S. government agencies, library organizations, and institutional repositories. The U.S. public domain provides an opportunity to increase the number of peer-reviewed journal articles that are open access.
Responsible Research Evaluation 2024: Summary of the SCOPE Workshop
Wolf, Baron G.; Miles, Rachel A. (NCURA, 2024-08)
Over the past six years, the INORMS Research Evaluation Working Group has worked closely with groups across the academic sector to help them use the SCOPE Framework in their own research evaluation exercises. In 2023, the Institute of Museum and Library Services (IMLS) awarded a grant (grant number: LG-254850-OLS-23) to Dr. Baron Wolf, Assistant Vice President for Research and Director for Research Analytics at the University of Kentucky, and Co-PI Rachel Miles, Research Impact Coordinator at Virginia Tech University Libraries, to workshop the SCOPE Framework in the US at a two-day, in-person forum that brought together librarians, researchers, university administrators, and research managers and provided formal training in making strategic decisions using research evaluation methods. The forum took place in Albuquerque, New Mexico, March 13-14, 2024 (https://evaluationforum.uky.edu).
Widespread exposure to SARS-CoV-2 in wildlife communities
Goldberg, Amanda R.; Langwig, Kate E.; Brown, Katherine L.; Marano, Jeffrey M.; Rai, Pallavi; King, Kelsie M.; Sharp, Amanda K.; Ceci, Alessandro; Kailing, Christopher D.; Kailing, Macy J.; Briggs, Russell; Urbano, Matthew G.; Roby, Clinton; Brown, Anne M.; Weger-Lucarelli, James; Finkielstein, Carla V.; Hoyt, Joseph R. (Springer, 2024-07-29)
Pervasive SARS-CoV-2 infections in humans have led to multiple transmission events to animals. While SARS-CoV-2 has a potential broad wildlife host range, most documented infections have been in captive animals and a single wildlife species, the white-tailed deer. The full extent of SARS-CoV-2 exposure among wildlife communities and the factors that influence wildlife transmission risk remain unknown. We sampled 23 species of wildlife for SARS-CoV-2 and examined the effects of urbanization and human use on seropositivity. Here, we document positive detections of SARS-CoV-2 RNA in six species, including the deer mouse, Virginia opossum, raccoon, groundhog, Eastern cottontail, and Eastern red bat between May 2022–September 2023 across Virginia and Washington, D.C., USA. In addition, we found that sites with high human activity had three times higher seroprevalence than low human-use areas. We obtained SARS-CoV-2 genomic sequences from nine individuals of six species which were assigned to seven Pango lineages of the Omicron variant. The close match to variants circulating in humans at the time suggests at least seven recent human-to-animal transmission events. Our data support that exposure to SARS-CoV-2 has been widespread in wildlife communities and suggests that areas with high human activity may serve as points of contact for cross-species transmission.
Introducing AI for LAMs: A Beginner Tutorial for Practical Generative AI Use Cases
Chen, Yinlin (2023-11-15)
Generative AI and Large Language Models (LLMs) are transforming various fields, including libraries, archives, and museums (LAMs). This workshop is specifically designed to introduce LAM professionals to the fundamentals of Generative AI and LLMs, utilizing hands-on applications through platforms and frameworks like OpenAI API, Hugging Face, LangChain, and more. Participants will benefit from practical exercises and tutorials, as well as an in-depth demonstration of selected real-world projects that underscore the transformative potential of AI. Moreover, the workshop will include a focused discussion session to foster brainstorming on strategies, methodologies, and the challenges of seamlessly integrating AI into traditional LAM environments. Guided by a University Libraries professor experienced in teaching "Introduction to Artificial Intelligence" in Computer Science courses to over five hundred students, this half-day workshop offers a blend of academic insight and practical expertise. Participants will gain hands-on experience with AI tools, learning to apply these emerging technologies creatively and efficiently. Tailored to LAM professionals curious about AI and its potential applications, the session serves as an insightful introduction and a comprehensive guide for those eager to augment services within the LAM settings.
Health Worker Potential for Expanded Exploration of Public “Frontlineness”: A Scientometric Analysis
Bredenkamp, David M.; Abdelrasol, Saif Tarek; Boyette, Charity L.; Comer, C. Cozette; Stovall, Connie; Talukdar, Shahidur Rashid (2024-06-28)
Public-sector frontline service scholarship in the field of public administration has been conducted under relatively limited circumstances and contexts. While literature focusing on the topic has been prolific, the context and lenses through which “frontlineness” has been viewed and observed are more limited (Chang & Brewer 2022). The scholarship on street-level bureaucrats (SLBs) has focused on a well-defined, though narrow, set of workers and work environments (e.g., teachers and nurses; schools and hospitals); those concentrated and consistent parameters may present an opportunity for greater generalizability of our understanding of SLBs than previously realized. We seek something of a new beginning: for theoretical exploration, clarity, and eventual reassessment of what frontlineness is and what it means. Healthcare has been a field in which public administration scholars have—either adjacently or directly—explored the nature of frontline work. We hypothesize, however, that there is much territory that goes unexplored due to siloing of disciplines, narrow definitions of what it means to be on the “frontline,” and more limited use in public administration scholarship of available evidence synthesis methods. One such method, scientometric analysis, provides useful tools to explore the potential of fields such as healthcare, with its results providing the “lay of the land” for further exploration. Using a scientometric analytical approach, this paper offers an answer to the following research question: What is the potential for existing research to describe the proximal relationship between a frontline healthcare employee and the frontline itself?
LLMs for Semantic Web Query
Chen, Yinlin (2023-11-09)
The emergence of Large Language Models like GPT-4 offers unprecedented capabilities in understanding human intent and generating text. This tutorial explores the intersection of LLMs and semantic web applications, focusing on how these models can automatically generate queries that adhere to metadata standards. Participants will engage in hands-on exercises that demonstrate the integration of LLMs into a sample semantic web application. This session will offer conceptual understanding and practical skills for metadata practitioners, developers, and researchers. The aim is to enable attendees to leverage the capabilities of LLMs in enhancing semantic web applications. Target audience: Metadata practitioners, developers, researchers, and those interested in Large Language Models Expected learning outcomes: Understand LLMs and their capabilities. Gain hands-on experience and learn to generate metadata-compliant queries using LLMs. Discuss potential applications and limitations of LLMs in the semantic web. Tutorial style: Presentation, demonstration, hands-on practice, discussion and Q&A Prior knowledge required: Basic familiarity with semantic web technologies, such as RDF or SPARQL Some basic Python programming skills Participants are recommended to have: A dual-monitor setup or two computers to more easily follow along with hands-on exercises while also watching the presentation
Streamlining DBpedia Queries with Natural Language Using Large Language Models
Chen, Yinlin (2023-07-10)
The capability to query knowledge bases like DBpedia using natural language is an emerging approach in the semantic web and linked data. This presentation highlights the use of GPT, a large language model (LLM), to examine its potential for interpreting natural language queries and retrieving information from linked data repositories. Think of the convenience of querying DBpedia with questions such as "Where was Albert Einstein born?" or "Who won the Nobel Prize in Literature?". To retrieve such information today, one must understand and write SPARQL queries. LLMs, like GPT-4, have the potential to translate these natural language queries into SPARQL, thereby making DBpedia more accessible to those without technical expertise in SPARQL. This approach improves the search experience and paves the way for more intuitive interaction with linked data. While there are challenges to this approach, including ensuring the accuracy of generated SPARQL queries and handling ambiguous natural language inputs, the integration of GPT-4 with DBpedia opens up a new avenue in information retrieval. This presentation will explore this promising approach, demonstrating its potential to modify our interaction with linked data and influence its practical use in the future.
Transforming Libraries for the Future: Elevating Service Innovation with Generative Artificial Intelligence and Prompt Engineering
Chen, Yinlin (2024-03-25)
This presentation introduces a specialized prompt library for academic libraries, focusing on collecting and sharing effective prompts to fine-tune Large Language Models (LLMs). By gathering a wide range of prompts that reflect real-world academic queries and scenarios, this initiative seeks to enhance the adaptability and accuracy of LLMs. The prompt library acts as a vital instructional dataset for AI training and fosters a collaborative environment where librarians, educators, and researchers can exchange effective prompts and develop new techniques. This project underscores the role of bridging digital libraries, information science, and AI. It showcases how prompt engineering and generative AI can transform academic libraries into more responsive, efficient, and intelligent resources, thereby contributing to the broader communities.
Librarian-in-the-Loop Deep Learning to Curate Very Large Biomedical Image Datasets
Xie, Zhiwu; Chen, Yinlin (2024-02-01)
We present a research data management project where librarians from University of California, Riverside and Virginia Tech are deeply embedded in a research team at Yale School of Medicine to directly answer specific research questions by applying AI/Deep Learning techniques to very large biomedical images. Leveraging library resources and expertise, we have developed a prototype pipeline that identifies nuclear pores from whole cell images captured at 8 nanometer resolution by a cutting edge microscope, in the hope to reveal the cellular mechanism of one type of epilepsy and autism. This project exemplifies out data management approach that strives to engage in much earlier stages of research, e.g., even during ideation and data collection, instead of waiting till most research activities are completed to "consult" or "advice" on the very general questions on data storage or preservation. This project also highlights the importance of non generative AI approaches, which have already been widely used as research tools in a much more mature manner.
Building a full-stack Serverless Web application with React and AWS
Chen, Yinlin (2021-06-10)
Serverless computing allows you to build Web applications without managing or maintaining servers. Using AWS, we can build and deploy responsive applications in the cloud with built-in high availability and flexible scaling capabilities. In this workshop, we will learn how to build a full-stack serverless Web application using React and several AWS services, including AWS Amplify, Lambda, AppSync, DynamoDB, etc. We’ll start the workshop with a quick overview of serverless computing and AWS, followed by creating a React application, integrating with AWS managed services and deploying this application in AWS. Workshop Agenda: Introduction to AWS, Serverless, AWS Amplify, and React Section 1: Create your first React application and setup AWS Amplify Section 2: Setup access controls for your application Section 3: Introduction to GraphQL and AWS AppSync Section 4: Perform data mutations for your application Section 5: Introduction to multiple development environments; Wrap-up and discussion.
Code4AI
Chen, Yinlin (2023-03-14)
Telling the Extension Story: How to Tell a Good Story for Connection & Advocacy
Haugen, Inga (2024-04-25)
Storytelling for Connection and Advocacy, telling your personal and your professional Extension stories.
Social network interventions to reduce race disparities in living kidney donation: Design and rationale of the friends and family of kidney transplant patients study (FFKTPS)
Daw, Jonathan; Verdery, Ashton M.; Ortiz, Selena E.; Reed, Rhiannon Deierhoi; Locke, Jayme E.; Redfield III, Robert R.; Kloda, David; Liu, Michel; Mentsch, Heather; Sawinski, Deirdre; Aguilar, Diego; Porter, Nathaniel D.; Roberts, Mary K.; McIntyre, Katie; Reese, Peter P. (Wiley, 2023-07-03)
Introduction: Racial/ethnic disparities in living donor kidney transplantation (LDKT) are a persistent challenge. Although nearly all directed donations are from members of patients’ social networks, little is known about which social network members take steps toward living kidney donation, which do not, and what mechanisms contribute to racial/ethnic LDKT disparities. Methods: We describe the design and rationale of the Friends and Family of Kidney Transplant Patients Study, a factorial experimental fielding two interventions designed to promote LKD discussions. Participants are kidney transplant candidates at two centers who are interviewed and delivered an intervention by trained center research coordinators. The search intervention advises patients on which social network members are most likely to be LKD contraindication-free; the script intervention advises patients on how to initiate effective LKD discussions. Participants are randomized into four conditions: no intervention, search only, script only, or both search and script. Patients also complete a survey and optionally provide social network member contact information so they can be surveyed directly. This study will seek to enroll 200 transplant candidates. The primary outcome is LDKT receipt. Secondary outcomes include live donor screening and medical evaluations and outcomes. Tertiary outcomes include LDKT self-efficacy, concerns, knowledge, and willingness, measured before and after the interventions. Conclusion: This study will assess the effectiveness of two interventions to promote LKD and ameliorate Black-White disparities. It will also collect unprecedented information on transplant candidates’ social network members, enabling future work to address network member structural barriers to LKD.

Browse

Recent Submissions