Digital repositories shape how knowledge is organized and accessed, yet their metadata often reproduces linguistic hierarchies. This study analyzes queer and gender-diverse representation in India's National Digital Library of India (NDLI) and Shodhganga through multilingual keyword searches. Results show English terms like queer, LGBT, and transgender dominate while vernacular South Asian identity terms such as hijra, kothi, aravani, jogappa and mangalamukhi appear far less frequently within repository metadata fields such as titles, abstracts and keywords. To address these inequities, the study proposes the "Queer Indic Term Map" (QITM), a framework linking Indian vernacular terms with inclusive vocabularies like "Homosaurus" and "Gender, Sex and Sexual Orientation" (GSSO). QITM demonstrates how culturally grounded metadata can enhance inclusivity and visibility in national digital repositories.
Faceted tradition occupies a foundational role in Knowledge Organization, yet the persistent conflation between faceted analysis and faceted classification has obscured its epistemic scope and methodological depth. This article departs from the hypothesis that distinguishing faceting as a logical-analytical process (faceted analysis) and as a formal representational structure (faceted classification) is essential to prevent its reduction to a merely technical mechanism. Through a historical-conceptual reconstruction of Ranganathan and the Classification Research Group (CRG), we demonstrate that these two instances operate at distinct levels of abstraction and formalization, fulfilling complementary but non-interchangeable roles. Our findings indicate that the enduring power of the faceted approach lies not in the classificatory artifact alone, but in the precise articulation between method and product, which is indispensable for designing semantically expressive, interoperable, and context-adaptive information systems. We conclude that reclaiming faceting as an epistemological paradigm rather than a technical resource is key to repositioning it within current debates on ontologies, folksonomies, digital infrastructures, and artificial intelligence, by clarifying the epistemological distinction between analytical method and classificatory product.
This case study examines the application of generative artificial intelligence (AI) tools, specifically ChatGPT-4o, ChatGPT-5, and Google Gemini 2.5 Flash, to streamline metadata extraction and creation for a born-digital collection within a digital repository environment. Testing four descriptive metadata elements-titles, abstracts, keywords, and full-text URLs-the study shows that AI improves efficiency in structured tasks but remains limited in semantic ones such as keyword matching. Results demonstrate that AI can streamline metadata workflows for small or resource-constrained institutions while emphasizing the continuing importance of human oversight and ethical responsibility in AI-assisted metadata workflows.
Libraries may need to adopt track-based music organization as it is mainstream in online environments. This study investigates whether the current Work-Expression model could be applied to jazz track organization by collecting and analyzing jazz enthusiasts' opinions. Results show that improvisation and freedom are the two most agreed characteristics of jazz, and jazz enthusiasts tend to organize tracks of the same tune based on genre and style, and do not want the tracks pre-grouped but tagged with attributes such as instrumentation. It is concluded that the Work-Expression model is applicable to jazz when the tune is treated as the Work.
This study examines the state of inclusive metadata practices in academic digital libraries, focusing on R1 institutions in the United States. Through nine semi-structured interviews with metadata professionals across diverse regions and institution types, the authors explore how inclusive description is understood and implemented in digital libraries. Findings reveal widespread interest in inclusive metadata, but implementation is inconsistent, constrained by project-based approaches shaped by legacy data, system limitations, and inadequate resources. The study highlights gaps in shared standards, metrics, and cross-specialty coordination, underscoring the need for unified frameworks and further research tailored to digital library environments.
This study examines large-scale reconciliation of BIBFRAME Work and Hub entities through the matching of over 9,000 MARC records against Library of Congress BIBFRAME datasets. Results show wide variation in matching accuracy, low Hub matching rates, and predominance of one-to-one Work-Instance relationships. Key challenges include difficulties in identifying Hub entities and limitations of current reconciliation workflows. Manual evaluation indicates that reconciliation accuracy is shaped not only by matching algorithms but also by legacy MARC data quality. These findings highlight the need for clearer modeling guidelines, improved data quality, and sufficient computational resources to support reconciliation at scale.
This study evaluates ChatGPT's potential to generate accurate, efficient, and standards-compliant MARC 21 records for Arabic materials. It aims to streamline Arabic bibliographic cataloging workflows and enhance library operational efficiency. The methodology involves prompting ChatGPT to create records for a representative sample of Arabic materials, followed by a comparative analysis against human-generated MARC 21 records to identify any discrepancies in adherence to MARC 21 cataloging rules. The findings contribute to discussions on ongoing advancements in generative AI technologies in bibliographic metadata generation and cataloging standards compliance, and their influence on future models and frameworks in library and information science.
This paper explores the ideological underpinnings of the Library of Congress subject heading "Conduct of Life" in catalog records where it is used as a subdivision in relation to queerness. Using Foucauldian discourse analysis, we examine this subdivision and its collocation in the context of historical evolutions of "Gay people" as a "Classes of persons," revealing moralistic judgments regarding queerness in the use of Library of Congress Subject Headings (LCSH). We argue that this subdivision work identifies queer embodiment as morally divergent and discuss how ideologically bound concepts trouble notions of 'ofness' and 'aboutness' while reinforcing issues of essentialism with Library of Congress subject headings. The paper concludes with theoretical implications and practical approaches to redress marginalizing subjectivity in description.
This article discusses the state of linked data (LD) for resource description among information professionals, end users, and third parties in the Galleries, Libraries, Archives, and Museums (GLAM) sector. It is based on an online survey conducted by the authors, which takes stock of current LD implementations, including technologies, training, and staffing levels. Key findings include skepticism among North American GLAM professionals about LD's value proposition and the key roles of training, open-source software, and strategic partnerships. The findings indicate that for many professionals in North America, LD is stagnating in a state of perpetual promise.
This article reports experiments with learning to rank (LTR) in the public library context. We used training labels based on human relevance judgments of query-document pairs, along with combinations of features drawn from catalog data, to train ranking models using two algorithms. We evaluated the models by measuring the ranking shifts of chosen hits. The results indicate that LTR is a method with much potential to improve the ranking of materials in a public library catalog.
The Oslo public library (Deichman) offers search functionality to its catalog over Elasticsearch. For interested patrons, it offers control over two reranking parameters, AgeGain and ItemGain, that allow for boosting the influence of freshness and popularity, respectively, for individual searches. In this paper, we explore the effect of these parameters on several retrieval scenarios and their effect on the quality of retrieval in the Deichman case. We find that whereas AgeGain has little effect on the search result, high values of ItemGain boost known-Item search but may have undesired consequences for topical search, as expressed through traditional retrieval measures.
