A project based out of U of Michigan works to "harvest" publically accessible (but often hidden or hard to find) digital resources for research from across the Internet. I haven't had time to play with it much yet, but it looks like a great resource (and not just for academics). OAIster Project Description:
"The University of Michigan Library service establishes a broad, generic, information retrieval resource for information about publicly available digital library resources provided by the research library community. This service is built through a collaboration that relies on the University of Illinois's metadata harvester.
The two primary criteria for inclusion in the University of Michigan service are that the information resources the metadata describe are:
publicly accessible and have no access restrictions, and
have a corresponding web-based digital representation (e.g., this would not include the metadata records for slides when the slides cannot be accessed through the web).
The service also encompasses as broad a collection of resources as possible (i.e., with no subject parameters). The service is accessible to the entire Internet community, without bounds. We hope to finally begin to reveal these 'hidden web' digital library resources in a way that they are not now revealed. "