Automating the Maintenance of Greenstone Collections

Abstract

Many applications generate multimedia documents, such as images and video, on a daily basis. For example, many municipalities have a photo-radar system to catch vehicles that violate traffic laws such as speeding or ignoring red lights. Many pictures of vehicle license plates are created every day. If these images are organized into a digital library, the collection would need to be updated regularly to incorporate new images. Another example is a traveller who wants to update a digital library of trip photos with new pictures of her travels while traveling around the world. When documents and metadata are added to a digital library collection on a regular basis, such as hourly, daily or weekly, an automated and scheduled approach to collection maintenance is preferred over having to manually update the collection. In addition, an automated approach should be simple to use, therefore making time available for other important tasks. Digital library software such as Greenstone (Witten et al., 2009), DSpace (Tansley et al., 2006) and Fedora (Lagoze et al., 2006) require that items be added manually to the collection. In Fedora, data is retrieved at the time of viewing. However, a location needs to be manually configured. Further, although Fedora and DSpace do provide application programming interfaces (APIs) to extend functionality, programming knowledge is required for using an API and for setting up tools based on it. We present a solution for automating and scheduling updates that occur on a regular basis. Our solution is implemented in the Greenstone digital library software system (Witten et al., 2009), and comes in two parts. The first part of our approach is a command-line scheduling module. The Scheduler both automates the construction and modification of a collection, and schedules the construction to occur at specific intervals, such as hourly, daily or weekly. In addition, the owner of a collection can update the collection manually, without affecting the scheduled collection builds. Further, the Scheduler interacts with the existing task scheduling mechanism on the host system, which keeps the Scheduler minimal, yet powerful. The second part of our approach involves incorporating the Scheduler into the Greenstone Librarian Interface (GLI) (Witten, 2004). This will allow users who are more comfortable with managing collections through a graphical user interface to take advantage