Case Study. British Library 19th Century Book Digitisation Project

Transcription

1 Case Study British Library 19th Century Book Digitisation Project

2 I. Introduction 1. About the British Library The British Library is the national library of the United Kingdom. It holds over 150 million items in many different languages and formats. As the main legal deposit library it receives copies of all publications in the United Kingdom and Ireland. The current organisation was formed by administratively combining a number of separate library departments to form the British Library bringing the Library into operation with effect from 1 July Today the British Library (BL) is one of the most important libraries in the world. 2. Goal and scope of the project The scope of the project was to digitise up to 25 million pages of 19 th century books selected by the British Library and Microsoft. The digitisation took place at the main British Library building in London St Pancras from 2007 to The project was funded by Microsoft within their Book Digitization Program. The British Library was permitted to use the digitised items for a wide range of purposes. In addition to widening access, the project also contributed to the long term preservation of the 19 th century material. The British Library received METS/ALTO and JP2 files for long term preservation. CCS applied the process on-site, British Library. integrating with the existing infrastructure and workflows at the 3. Challenges very limited time frame for this large scale project; fragile books which required careful handling; immediate and constantt quality control of the digital objects (image quality, OCR quality, completeness and metadata correctness, especially for multi-volumes); data consistency of the final output; large data volume required immense IT environment; staff and production planning to maximise resources. 4. Executive Summary After the period of 20 months the agreed target of the 23,000,000 pages was achieved and the project successfully completed. Two project managers, Neil Fitzgerald from British Library and Joachim Bauer from CCS, and 10 scan operators worked in London. Scanning was performed using four Kirtas APT 2400 scan robots and one Zeutschel 0S overview scanner. Each scanner was 2 / 8

3 always attended by one scan operator in order to check the scanning results in docworks ScanClient. While the Kirtas scanners were used for high speed scanning of regular pages, the Zeutschel scanner was used for binding and foldout scanning. This workflow accelerated the overalll performance significantly and was supported by docworks status handling. The scanned pages went throughh the docworks digitisation workflow: Cropping, zoning, page sequencee and metadata were verified and corrected in docworks at CCS s offshore location in Bucharest, Romania. The output format was METS/ALTO and JP2 with dynamic compression (peak signal to noise ratio). The British Library performed immediate and constant quality control of the digital objects and the data was ingested into the British Library's digital library system right from the start. This was essential for quality verification. Number of monthly digitised pages between April 2007 and December 2008: II. Environment 1. Team 2 project managers (British Library, CCS) 2 studio managers in two shifts (CCS) 10 scan operators in two shifts (CCS) 2 book handling staff (British Library) 2 IT staff (British Library, CCS) 60 docworks operators in Romania (CCS) 3 / 8

4 2. IT environment / hardware: The digitisation process (except the manual QA) was performed at the British Library, St Pancras. The scanning studio was in a dedicated area while two full racks with 40 TB storage (fibre channel and SATA) and 14 IBM blade servers were incorporated in the British Library data centre elsewhere. A redundant 10 Gb uplink ensured fast upload from the scanners to the storage. Scanning equipment: 1 Zeutschel OS and 4 Kirtas APT 2400 scan robots, 5 book tracking. A 1 Mb internet connection was Romania. barcode readers for available for 24/7 data transfer to and from the offshore location in 3. Scanning studio The scanning studio at the British Library provided sufficient room for the actual scanning operation and book storage for a rolling stock operation. 4 / 8

5 4. Software docworks digitisation software (including Abbyy Finereader) docworks ScanClient ItemTracking SQL database MS Access MS Sharepoint III. How did they do it? 1. Project Management / Project Phases Pilot phase: from April 2007 to June 2007 System design Workflow design Establish on-site digitisation studio Align process with existing infrastructure and workflow at BL Setup of IT infrastructure (scanners, network, storage, processing units) Installation of one Kirtas scan robot and the Zeutschel scanner The CCS project manager trained the first two scanning operators, defining processes and creating the documentation Interaction with conservation team for agreement of book handling procedures Book handling training by British Library conservation department Verification of scanning and output quality Established automated data transfer between Automatic Book Retrieval System (ABRS) and CCS Workflow database Integration of metadata records into docworks workflow Setup of scalable book transfer workflow for mass digitisation Developed and synchronised digital object delivery procedures QA feedback and knowledge transfer Ramp up phase: From July 2007 to September 2007 Increase of the throughput up to 1,000,000 pages per month Anchoring of the developed structure Set up of two more Kirtas scanners Training six more scanning operators Establishing studio manager position Production phase: From October 2007 to April 2009 Monthly throughput of about 1.5 million pages Adding one more Kirtas scanner 5 / 8

6 Training two more operators 2. Overall Workflow The BL project team chose the items to be digitised, using the ABRS to create approximately 250 single item requests which were grouped into one batch. The system was used to print barcode slips including a unique ID, the title and the shelf mark for each request. Obviously the selection was made for complete shelves whenever possible for collection management purposes. The library staff were responsiblee for book movements until handover to the CCS studio manager. One of the most important actions in the workflow was the book in procedure. The studio manager scanned all barcodes in order to verify and record that the books had entered the studio. Later in the process every status change (scanning started, scanning complete, foldout scan, cover scan, discussion, book out) was recorded in the same way. This was essential for the project s success as the various stakeholders demanded an extremely high level of transparency. 6 / 8

7 The studio manager arranged the books by condition and size in order to reduce scanners. set up time at the CCS developed an item tracking system which supported the workflow and enabled an automated delivery procedure and overview. This system used an SQL database MS Access interface and MS Sharepoint layer for delivery reporting. 3. Technical details (use of docworks) a. QA manual work The Remote Quality Assurance took place at CCS Romania. 60 operators used the docworks software for manual corrections and verification. Manual correction / verification steps: Verify Page Frames (Cropping) Verify Layout Elements (Zoning) Verify Page Numbers (Page sequence) in one step Additional manual correction / verification step for multi-volumes: VPH (assigning proper metadata record) 7 / 8

8 b. Output formats For each book a digital object was created including the processed images, XML data including the OCR text, page sequence numbers, and information about illustrations, technical and descriptive metadata. This output data was automatically validated using XML validation and checksums. Dynamically compressed JP2 file as master image ALTO XML files per image representing layout and text information complying to ALTO schema ALTO-1-3.xsd METS.XML file containing MARC descriptive meta data, METS rights copyright meta data, MIX technical meta data, filesec containing information about all files and sequence, physical structmap containing information about physical structure, page sequence and page type (title page, table of content page, foldout page) and a logical structure containing information about the sequence of text and illustrations METS SHA file contains hash code SHA-1 for METS file for validation that the file has not been changed Bound PDF file containing all images as dynamically compressed JP2 files, with hidden text, linearised to open on the table of content page, title page or first image of the book if both are not available. The PDF file contains certain meta data (if available from catalogue record) as properties (title, author, keywords). IV. 19 th century books today When Microsoft wound down their book digitisation programme the British Library explored new ways to make the content available. In 2011 the British Library teamed up with BiblioLabs, a US historical content reseller based in Charleston and made content available in the British Library 19 th Century Collection App: and via the JISC platform: for example. V. Client s statement / contact details CCS has been contracted by Microsoft to digitize 100, th century books at the British Library. We are extremely satisfied with the approach, implementation and delivery of this project by CCS. The BL has no hesitation in recommending them for similar work at other institutions/organizations. (Richard Boulderstone, Director e-strategy and Information Systems at the British Library) Contact Person: Neil Fitzgerald (External Technical Projects Delivery Manager at the BL) Neil.Fitzgerald@bl.uk 8 / 8