Quantity meets quality: towards a digital library

Abstract

In 2006 the Koninklijke Bibliotheek (National Library of the Netherlands) set up a production chain to digitize 8 million newspaper pages within five years. The scale of the project meant moving from state-of-the-art digitization towards a workflow where quality needs to be balanced against quantity. A completely new workflow for mass-digitization, from selecting titles to implementing scans in the KB infrastructure and publishing digital newspapers online, had to be implemented by a heritage institution that - after nearly 15 years of working with digitization - is rapidly evolving into a true digital library. As part of this process, digitization and OCR is outsourced. Newspaper bindings are delivered to and collected from the library once a week, using special transportation boxes. The location of the physical documents and the digitization process are mapped using a barcode tracking system that also enables quality and completeness checks to be performed. In preparation for article separation. the newspapers are scanned in full colour. Layout, text and structure elements are recognized and meta data collected. The documents are then issued in METSIALTO, MPEG2 1, JPEG 2000 and PDF formats, which undergo extensive quality checks before they are accepted. To process the results, the KB built online services on a new infrastructure using open source software and standard protocols (SRU, OAI, Dublin Core) to ensure accessibility. The decision to develop online services in-house means that managing and maintaining the digital newspaper databank has become part of the existing organization. This presents a huge challenge, not only for the technical infrastructure but also for the KB as a whole in terms of staff and workflow management. The KB and its supplier would like share with IFLA members valuable best practice in creating a digital newspaper library.