Keynote 2: Maintaining Metadata Simplicity

In the modern world of knowledge institutions, we strive to describe a wide range of settings. The contexts and use cases of academia are different from those of local libraries, elementary schools and exploratory research. Some scenarios require detailed data, others call for broad overviews. The metadata produced and used in these settings are varied and diverse, forming descriptions of differing dimensions.

The National Library of Sweden serves the Swedish cooperative union catalog (Libris), which has different audiences both nationally and internationally. To overcome the silo effect of old technology, and to interoperate with different metadata standards, we have developed a catalog system based on RDF, using linked vocabularies and datasets. As we gain production experience with this system, we strive to reduce complexity coming from decades of data collection and the evolution of multiple interdependent systems, both internal and external to our organization.

It can be hard to harmonize disparate metadata into coherent, interconnected and reusable descriptions. As an example, the set of Dublin Core metadata terms have been central in the evolving practices of structured descriptions since the turn of the century. They have been used in many ways, from making web documents searchable to enabling navigation of interlinked knowledge graphs. They have an essence of simplicity, enabling information retrieval and interoperability through their flat and general style. But in today’s landscape of diverse metadata management, this set of terms is only one among many different description models to choose from.

There is much value in keeping this essence of simplicity when mapping and navigating the depth and complexities of our growing knowledge base. We are working on practical ways to bridge differences between various models, by consolidating similar terms and reducing overlapping information. Our vocabulary maps to Dublin Core, Schema.org and BIBFRAME, and our descriptions are interlinked with LCSH, RDA, Wikidata and beyond. We aim for an application design exposing as much or as little as necessary for different settings, while maintaining connected graphs with shared identifiers and common definitions.