Knowledge Organization Systems in the Age of Artificial Intelligence: Challenges for Library Cataloguing

Abstract

The emergence of large language models and AI-driven classification systems presents both opportunities and fundamental challenges for traditional library knowledge organization. This paper critically evaluates the application of automated subject classification to library cataloguing workflows, drawing on implementation data from 12 public library systems that piloted AI-assisted cataloguing tools between 2022 and 2024. While AI systems demonstrated 87% accuracy on standard subject headings, performance degraded significantly for interdisciplinary works and materials in non-dominant languages. We argue that effective knowledge organization in the AI era requires hybrid human-machine workflows that preserve cataloguer expertise while leveraging computational efficiency. Implications for Library of Congress Subject Headings, Dewey Decimal Classification, and linked data initiatives are discussed.