Friday, December 13, 2002


In October 2002, members of the core integration team for the National Science Digital Library (NSDL) program issued a report on progress being made in the continuing effort to improve metadata interoperability in harvesting digital library collections. NSDL's ultimate goal is to integrate the tens of thousands of collections, ranging from simple Web sites to large and sophisticated digital libraries, into a coherent whole that is structured to support education and facilitate incorporation of innovative, value-adding services. Because NSDL can only coax and cajole collections toward preferred standards, their harvesting architecture needs to accommodate a wide spectrum of interoperability, which makes use of widely varying protocols, formats and metadata standards. In designing the architecture, therefore, NSDL's Core Integration team explicitly recognized the necessity to accept whatever metadata the collections can provide, which in many cases is very basic, in any of several preferred metadata formats. The team creates the collection-level metadata, but not the item-level records. To provide a minimally uniform level of metadata for all known items, in addition to storing the native metadata (provided by the collections), a Dublin Core record is created for each item in a format called nsdldc. This format contains a number of Dublin Core elements, including some elements unique to the NSDL, but the actual record may be little more than a simple identifier. While the NSDL Core Team believes it would be too optimistic to hope that every collection will support its metadata harvesting program, the increasing positive responses from many important collections would seem to predict that the program will achieve the desired level of popularity. (Cornell University Oct 2002)

From ShelfLife, No. 84 (12 December 2002) ISSN 1538-4284