Putting meat on the bones: describing image collections (without any staff!)

Abstract

The University of Edinburgh’s "Library and University Collections" is very proud of its high-resolution images of the wealth of Special Collections it holds. The discovery of these images is handled by the LUNA Imaging platform, a supplied system, which allows high quality JP2K zooming, and also presents its metadata using robust solr indices. Getting the data into the application has presented us with a number of interesting challenges. To briefly describe this workflow: our Photographers receive readers’ orders from items they have found in our manuscripts, and these are recorded using Excel worksheets (we have offered to move the whole process to the web but for various reasons, this has not happened!); we take the shorthand data that they record and turn it into presentation standard, using an Excel macro which features various programming techniques; this macro also runs file renames, and runs a process to embed identification data into the TIFFs. From here, per collection CSVs are generated for upload to the system, which parses the particular CSV into the relevant format under the covers; this gives us a skeleton record in the LUNA system. As we do not have cataloguers devoted to our images, we need to be creative to enrich the records to make them searchable. We have built a purpose-built crowdsourcing application based on standard LAMP technologies to allow the crowd to further catalogue the record. The data is then hived off to the correct standard using JSON or XML, and run back in using processes we’ve built around the system’s REST API. This end-to-end workflow has grown organically, does everything we need it to do, and has CSV at its very heart.