The automation of subject indexing and the role of metadata in times of Large Language Models

Abstract

So far, virtually every system that facilitates access to and exploration of information resources – library catalogues, discovery systems, research information systems – has metadata as a central component. Accordingly, generating and curating high-quality metadata has always been a core activity of information infrastructure institutions, especially libraries. This includes creating or extracting semantic metadata, also called subject indexing, i.e., the enrichment of metadata records for textual resources with descriptors from a standardized, controlled vocabulary. Due to the proliferation of digital documents, it is no longer possible to annotate every single document intellectually, which generates the need to explore the potentials of automation on every level.