TCDL 2026

As the bridge that connects users to digitized archival materials, descriptive metadata is an important tool for resource discovery and access. For this reason, metadata should be as consistent and accurate as possible; however, achieving and measuring quality metadata is no easy feat. In June 2025, the University of North Texas Libraries' Digital Projects Lab completed digitizing and describing the Rosenberg Library’s Harris and Eliza Kempner collection, made up of personal and professional materials from one of Galveston’s founding entrepreneurial families, for the Portal to Texas History. This project is seven years in the making and includes nearly 67,000 items, all described by over 50 different editors. Naturally, quality errors abound -- from typos that were never noticed to outright changes in metadata standards and name authority headings over the years. But how does one handle the analysis and improvement of a collection of this size? In this presentation, we will discuss a short-term project to iteratively improve metadata quality for the Kempner collection using the model outlined in Tarver et. al (2022) “EPIC: An Iterative Model for Metadata Improvement.” The EPIC model—Evaluate, Prioritize, Identify, Correct—breaks down the formidable process of metadata improvement into smaller, more easily-handled tasks. Through the use of an “epic” color-coded spreadsheet, we divided the tasks into tiers that allowed us to "level up" when a section was completed, providing a rewarding sense of accomplishment with each level completed. This model can be utilized by other institutions in their large digital collections to iteratively improve the quality of metadata and ensure easier access by users.