A Distributed Digital Library Collaboration
The American Memory collection The Nineteenth Century in Print: The Making of America in Books and Periodicals will be assembled in phases as a distributed digital collection through the collaboration of the Cornell University Library and the University of Michigan Library with the Library of Congress. The books and periodicals are drawn from the collections of the three institutions, many digitized in the original Making of America project, a collaboration between Cornell and the University of Michigan, with funding from The Andrew W. Mellon Foundation.
CollaborationThe first Making of America project began in 1995 as a collaboration between the libraries at Cornell and the University of Michigan. One primary objective was the assembly of a large, thematically coherent body of material that would serve research and teaching needs at the two universities. Another was the development of models for selection, conversion, storage, retrieval, and use of digitized materials on a large, distributed scale. Selection focused on published materials documenting American social history in the second half of the nineteenth century. The University of Michigan materials are primarily from the period 1850-77. The common approach to cost-effective conversion of run-of-the-mill printed text included use of high-resolution bitonal images (which can be compressed efficiently in a lossless fashion) and uncorrected optical character recognition (OCR). The University of Michigan developed a markup specification (an SGML DTD conforming to the guidelines of the Text Encoding Initiative) to represent the digital reproductions in a consistent and efficient form to support indexing and display. This markup specification has been adopted for use at Cornell. This project has generated over a million and a half pages, with high-resolution bitonal images and the corresponding searchable text. Cornell converted 910,500 pages and the University of Michigan 634,000 pages. In 1999, the Preservation Reformatting Division at the Library of Congress chose the same model for the conversion of the periodical Garden and Forest (8,400 pages); the OCR and markup for this periodical were performed for the Library of Congress by the University of Michigan Digital Library Production Service.
The two Making of America collections are available in their entirety through their parent institutions. Find them at Making of America (at Cornell University) and Making of America (at the University of Michigan). The first release (in early 2000) of The Nineteenth Century in Print integrated the books (monographs) digitized at the University of Michigan into American Memory and linked to the Library's first contribution, Garden and Forest. The second release (September 2000) incorporates the periodicals (serials) digitized at Cornell into American Memory. In May 2002, Garden and Forest was more fully incorporated into American Memory, supporting full-text searching in conjunction with the periodicals digitized by Cornell. This release also provided access to high-quality greyscale images of pages from Garden and Forest with printed halftone illustrations.
A Distributed Digital LibraryThe digital reproductions of the books and periodicals in this collection are mounted at different institutions, where they are managed and delivered on the World Wide Web through different software. The foundation for building digital libraries from content distributed among heterogeneous repositories managed by different institutions is interoperability. Interoperability can be attempted at many different levels. Any level of interoperability involves implementation costs for all parties, costs that must be balanced against the value of enhanced services supported by the combination of resources. Unfortunately, there is little experience to help institutions assess the value to users of providing different levels of interoperability to related resources that are already available independently. A very simple approach to interoperability has been used by the National Digital Library Program to integrate resources digitized and mounted by institutions winning awards in the LC/Ameritech competition into American Memory. For this new collaborative collection, the integration into American Memory will be based, at least initially, on the same simple level of interoperability. Copies of bibliographic records or the text derived from uncorrected OCR will be gathered at the Library of Congress and indexed along with the other resources within American Memory. When the user selects an item from a list of "hits" following a search or from a bibliographic display, the item will be retrieved from whichever institution holds the corresponding digital reproduction and displayed using the presentation style and navigation options provided by that institution.
The collaborating institutions will continue to build on the experience of this collaboration in many ways that will contribute to developing both the resource and shared experience with building large-scale distributed digital libraries. The University of Michigan Library has digitized roughly 7,000 more volumes from this time period to add to the Making of America resource. Cornell University Library is using its Making of America materials to compare designs for repositories and delivery systems for digital content. Cornell also plans to convert 10,000 nineteenth-century antislavery pamphlets to digital form. The two institutions plan to explore cost-effective open mechanisms for providing users with common navigation of the combined Making of America collections. The Library of Congress Preservation Reformatting Division will integrate the practices developed for its component of this project into the array of reformatting options available to support the preservation of materials in the Library of Congress collections. The Library of Congress will also integrate more of the content digitized in the Making of America project into American Memory as part of The Nineteenth Century in Print.