Reconstructing Provenance in Long-Lived Data Systems: The Challenge of Paradata Capture in Memory Institution Collection Databases

Paradata is important for understanding the provenance of data—but capturing and using paradata is challenging because it is often not formalized or explicit. This is particularly the case for complex, long-lived digital objects, such as the databases used to manage long-lived museum collections. These databases are passed down between generations of collections managers, but the documentation explaining their structure and changes over time is often incomplete, thus posing an obstacle to the use and maintenance of the databases. Collection managers must often reverse engineer their databases and create documentation from scratch. Here, we present a case study of paradata reconstruction conducted as part of a larger project studying database maintenance in memory institutions. Through interviews with collection managers at the University of Michigan Herbarium and Matthaei Botanical Gardens, we reconstruct how a database evolved and changed over 50 years. We show how different ways of illustrating the history of a database can be used to help “open up” a database for users. We reflect on the strengths and weaknesses of these approaches, specifically versioned entity relationship diagrams, Sankey diagrams, and narrative case summaries, and discuss the challenges in capturing paradata from long-lived sociotechnical objects.