As electronic libraries proliferate, containing vast databases of information and linked together by the information superhighway, we believe that distributed, standards-based, scalable electronic libraries are inevitable, and that they must support all current and future media in an easily accessible and content-addressable fashion. To place this global virtual library at the fingertips of a world-wide clientel will require the development of intelligent client programs that can aid the user in exploring the thousands of distributed information servers. It will also require application of advanced techniques for information retrieval, information filtering, resource discovery, and the application of new techniques for automatically analyzing and characterizing data sources ranging from texts to videos. We see a need to develop technologies to:
To this end, the Berkeley group has proposed a further research and development effort to continue the work begun on the CNRI CSTR project, under the NSF/NASA/DARPA Digital Libraries initiative. Electronic libraries are a fledgling technology with no firm standards, architectures, or even consensus notions of what they are and how they are to work. In the CSTR project at Berkeley, Stanford, CMU, Cornell, and MIT we are hoping to provide some of these standards, architectural models and definitions of the electronic library of the future.