With a grant from the Mellon Foundation, the University of Michigan Digital Library Production Services has developed OAIster, a “harvesting” engine for difficult-to-access digital resources, such as images, manuscripts, audio files and the like. These assets, in university, museum and library collections are often impossible to find using standard search engines which are unable to delve into the databases which store the assets and their metadata.
The Open Assets Initiative has developed a protocol for querying the cgi of these collections and capturing a great deal of metadata in XML format.
Many of the collections are managed with GNU EPrints, free software for creating online archives.