As linked data increasingly moves from experimentation toward implementation and production, GLAMs face a practical dilemma: how do we build scalable workflows while also ensuring that this data represents the collections and communities we seek to describe? Archival collections especially make this challenge visible. Their records often contain historically underrepresented creators that are inconsistently described or absent altogether from catalogs and authority systems. Linked data is aptly poised to be a lever for richer and more inclusive information access.
This presentation shares a developing linked data production workflow at the University of New Mexico using OCLC Meridian to create, connect, and enrich entities from archival collections. Beginning with MARC records and archival finding aids, the workflow identifies candidates for entity work, gathers disparate pieces of evidence, connects existing authority data, and models relationships. Experiments with OpenRefine, APIs, and Python explore how entity identification and reconciliation can scale while preserving human review.
This case study will share lessons learned about where automation helps, where it breaks down, and how gaps in existing descriptive infrastructure shape who, what, and where can become visible through linked data. The project argues that sustainable linked data workflows should consider not only technical scalability, but representation: using metadata expertise and local knowledge to build richer connections and ensure that underrepresented creators are present in emerging knowledge systems.
