This presentation shows the change in Duke University Special Collections' electronic records accessioning practice from 2007 to 2011. Presented both as part of the AIMS workshop (8/23/11) and at CurateGear (1/6/12).
This presentation briefly describes the evolution of accessioning electronic records within the Duke University Special Collections & Archives from 2007 until 2011.
When I arrived @ Duke in mid-2007 an accessioning process was already in place.
The process was documented in an 8-page manual that was difficult to understand and slightly different versions existed scattered through the shared network drive. This plus the still relatively in-frequent receipt of digital media created a situation where the Electronic Records archivist was always consulted for clarification & instruction. Additionally, the directions did not allow for fixity checking and could potentially result in modified date metadata.
The separation sheets asked archivists to list the files on each disk which was an unsustainable solution.
The DOS directory listing, while providing the date last modified, size, filename of the files, and a place for recording notes is not easily machine processable which could lead to later processing inefficiencies and lacked fixity information.
In early 2008 I embarked on a software project to simplify the accessioning process while adding fixity checking, generating machine processable metadata, and running additional tools such as Jhove and Droid. Also, separations sheets where now only required when necessary to retain original order of media found comingled with analog materials.
The Duke DataAccessioner prototype and initial internal release, built to automate the accessioning task, was completed within a week. Ease-of-use was the primary concern. The DataAccessioner provides the user with a single window to record a simple set of metadata, choose options, and initiate the data migration. When the user clicks “Migrate” the DataAccessioner creates a directory structure on the destination storage space which parallels the disk structure, generates checksums for each file on the source disk, copies them, verifies the checksum on the destination storage, and then runs the specified plugins. An XML based metadata record is generated and stored on the destination file-space.
By automating much of the work previously done by hand the 8-page guide was replaced with a two-page instruction manual which usually isn’t referred to after the first or second use.
3 years past with the DataAccessioner in production use with very few updates (improving the plugin functionality and a few bug-fixes). During this time I began experimenting with computer forensics and disk images resulting in the use of a hardware write-blocker for all accessioning (except FireWire drives which it did not support) and occasional disk image capture which were not run through the DataAccessioner. Also during that time Special Collections began using the Archivist’s Toolkit for recording accessions. In November 2010 Duke’s Office of Internal Audits audited the Library IT system to create a benchmark and see what improvements could be recommended. The materials, and more specifically, the potentially sensitive and un-controlled nature of the content took them by surprise. The discussions held by the audit spurred renewed attention to the existing accessioning workflow which was reaffirmed when the report was released in Feb 2011. The new revised workflow now uses barcodes to individually identify media, (usually) creates disk images, photographs the media, scans for SEI and viruses, and tracks the work in a workflow database (still in development) on a local machine before transferring to university storage systems. Our processes are never perfect. The key is to start somewhere and iteratively improve as the technology and circumstances allow.
A screen-shot of the current media tracking database.