Digital library developments currently emphasize traditional services, such as searching, archiving, and access to digital information (e.g., the HathiTrust and Google Book Search), but fail to recognize the potential for eScience to identify latent information that is buried in digital collections and scientific databases. This SGER is exploring the new forms of scientific research that are possible in an all-digital library. As a case study we are using Cornell's Mann Library, which will be one of the first large scientific libraries to digitize at least 80 percent of its book collections. As yet, only a few scientists are aware of the opportunities that an all-digital library offers. The first part of the study is gathering information from scientists about the potential for information-driven research in their domains, and the tools and services that they need. The second part of the study is undertaking pilot experiments to gain insights into the practical difficulties in carrying out information-intensive research on digital book collections. This will use an initial collection of 100,000 digitized books to which we are providing tools for indexing, simple data mining, and extraction of subsets.
Public dissemination of the results is an integral part of this study. The aim is not simply to explore eScience in the all-digital library, but to enable scientists to carry out eScience research without the help of computing specialists. For this they need support from universities, libraries, and the government. This support must be built on a well-informed analysis of the technical, human, and organizational factors that will maximize the benefits of the all-digital library.