NASIG Autumn 2024

The integration of AI into library practices offers significant potential for enhancing metadata quality. This session presents a project by the University of Toronto Libraries (UTL) that uses ChatGPT and Alma Cloud Apps to improve serial item metadata. The initiative employs AI-driven techniques to automate corrections and enhancements to item metadata and reload records using Alma's cloud apps. The project also demonstrates how these technologies enable retrospective itemization of serials and regeneration of summary holding statements. By utilizing AI and API tools, this scalable approach to metadata remediation allows for large-scale enhancements, improving the accessibility and utility of library collections.

1. Natural Language Processing and serials cataloging : an assisting perspective

What is under the umbrella of AI can really be a starting point in terms of assisting in serials description/cataloging? The complexity and different cases of serials nature, make it very difficult to rely on specific patterns that can be fed into an artificial intelligence model, in the context of teaching or training others on how to catalog a serial. On the other hand creating a trained chat bot on assisting in the basics of serials cataloging, with a serial expert, is something could be helpful. In this session I will present a chat bot I created and trained, for the purpose of the above.


2. Enhance! : Using Google Keep and Notepad++ to Automate Enhanced Contents Notes for a Large Scale Journal Cleanup

In summer of 2024, our cataloging department began work on a large-scale periodicals project requiring the recataloging, relabeling, and barcoding of roughly 12,000 volumes. To assist with the sheer quantity of materials, we cross-trained colleagues from other departments to manage the more straightforward journals, but this still left a large number of monographic series. To comply with our standards for bibliographic records, each item in these multi-author series would require enhanced 505 contents notes. By employing Google Keep for OCR and creating macros in Notepad++, we have streamlined this process to generate in five minutes what used to take twenty.

The Collection of the US House of Representatives began a Significance Assessment in 2023, with a goal of evaluating the collection, articulating and refining a collecting philosophy, and documenting institutional memory. This project revealed need for documentation beyond basic object research, and resulted in a great deal of new digitized collection information. This session will discuss our process, the challenges we encountered, and results so far in our efforts to make all this digitized research and information available, accessible and logically organized for users.

This case study details the University of Arkansas Libraries' initiative to restructure and flatten its library acquisitions fund model, reducing 370 funds and ledgers to under 85. This effort supports near-real-time financial insights via data dashboards within Alma Analytics and improves collection decision-making by enabling tracking of library expenditures in ways that better align with our collection development strategies. Extensive planning, stakeholder engagement, and a thorough review and updating of acquisitions policies and procedures were crucial to the project's success. The restructuring leverages iterative, project-based strategies, supported by detailed documentation, laying the groundwork for future acquisitions integrations with the campus procurement system.

Discover how Taskade, an AI-powered productivity platform, revolutionizes team and individual workflows. This session will showcase Taskade’s seamless integration of task organization, project specifications, workflow automation, and team collaboration into a single, intuitive platform. Learn how to effortlessly set up workspaces, manage tasks, and enhance teamwork for daily to-do lists or complex library projects. Join me to explore how Taskade can streamline your productivity and transform your work processes, no matter the size of your team.

This lightning talk will discuss a collaboration between a research group at the University of Maryland, Baltimore, and its Health Sciences and Human Services library to ingest ~300 reports into the University's digital repository. The reports and their descriptive metadata originally lived in a database designed by the research group. This talk will review the migration of the reports and metadata into the digital repository. It will primarily focus on the challenges around meeting the specific discovery needs of a research group with the broader parameters of an established digital repository.

Explore Goldey-Beacom College's inaugural Common Reading Program, featuring Kristen Radtke’s graphic novel, Seek You. Discover how students created visual expressions of the novel’s themes, blending traditional and AI-generated art. Learn about the integration of these works into the library’s digital asset management system, Quartex, and the implications for future digital archiving practices. This session will highlight the innovative use of AI in student artwork and the library's role in preserving these creations.

Among many electronic resources acquisition tasks, ensuring accurate titles for what libraries pay for involves comparing lists from publishers, previous years, and subscription agents. However, discrepancies within these lists pose a challenge, often requiring time-consuming manual comparison. To address this issue, the presenter envisioned developing a tool capable of efficiently comparing title lists and generating differential outputs. Crucially, they aimed to design the tool with user-friendliness in mind, making it accessible to colleagues with varying levels of technical expertise.
By harnessing the capabilities of ChatGPT/Copilot, they developed a user-friendly tool capable of automating reconciliation tasks. This tool simplifies the comparison process, enabling swift identification of differences between title lists. Its intuitive interface ensures ease of use for all users, regardless of their programming knowledge or experience.

Since the 1990s, pre-coordinated subject headings—long text strings comprising individual terms connected by double dashes (--) in the order of [topic] -- [place] -- [chronology] -- [form]—have been created in MARC records. Although these headings have been well-retained in the MARC environment, they introduce multiple discovery challenges in digital collections, where records are aggregated from both MARC and non-MARC environments. This session outlines the issues identified in the Florida Digital Digital Collections, details the workflow used to address them with MarcEdit and OpenRefine (two free data tools), and advocates for cataloging practices that consider needs beyond the original environment.

This presentation explores how AI can empower individuals without coding knowledge to write code, unlocking new opportunities and enhancing their ability to support their institutions. By leveraging AI for coding assistance, individuals can automate daily tasks, enhance collections, and develop solutions to simplify and expedite repetitive cataloging. This allows more time for improving metadata quality. Discover how AI can revolutionize your workflows and increase your impact within your organization.