METADATA ARCHAEOLOGY: AN IMPORTANT PART OF THE PROCESS OF DIGITIZATION OF PHYSICAL COLLECTIONS

Geological surveys, university departments, museums, and private collections are home to a vast store of geological physical specimens. However, oftentimes these collections are unknown outside of a select few. Recent efforts have focused on making these collections more discoverable by digitization and standardization of the metadata records associated with a sample. The Geological Survey of Alabama (GSA) has been collecting and storing geological samples since its founding in 1848. Expectations of records have changed significantly over this time. The various stewards of the collection have recorded information for a wide variety of specimen types, most often in their own system. Merging these various records has proved challenging. The metadata records are therefore often incomplete, and in some cases, completely lacking.

The GSA is undertaking what we have termed “metadata archaeology” to start to fill in the metadata blanks of these valuable resources. This involves sifting through old records (including formal and in-house publications, regulatory forms, and personal files), rediscovering information sources (“lost” records being rediscovered), and making personal contacts to find missing information and then evaluating what sources should be considered most reliable. This helps ensure that we have the most accurate metadata associated with our samples moving forward. A bit of serendipity is still needed to discover some of these records, but our sample collection is now moving towards 21^st century discoverability. The “more product less process” model, borrowing from the agile software development philosophy, encourages clear communication while encouraging simplicity. Quality is ensured through periodic checks with the original data provider (or their proxy) and/or verification through multiple sources. These workflows are meant to make the process of metadata rescue as straightforward as possible to help bring these potentially dark data to light.