Digital Library

Our practice of disseminating, accessing and using information, especially scholarly information, is still significantly impeded by the legacy of pre-electronic media. While overcoming these impediments will require many elements, there are opportunities for technological innovation to support new and better practices. For example, journals exist in their traditional forms at least partly because of the value of the peer review process, which thus far has not yielded to decentralized, distributed, and timely mechanisms of the Web. Similarly, information access is still largely a text-based affair, with other data types relegated to second-class citizenship.

The UC Berkeley Digital Library project is developing technologies aimed at addressing these impediments, and hence allowing the development of new, more efficient mechanisms of information dissemination and use. In particular, we are developing new models of documents, in the form of the Multivalent browser, which we hope will
convince you to throw away your current, limited Web browser, for "collaborative quality filtering," which provides the value of peer review without deference to prior established authorities, such as journals, and for "collection management services," which bring to individual information users services previously available to libraries. Taken together, such mechanisms may provide the benefits of modern communications without sacrificing traditional academic values.

In addition, we have been developing techniques for image retrieval based on image content. Recent progress on learning the semantics of image databases using text and pictures suggests that new forms of image-related Web services may be possible, including automatic image captioning and automatic illustration, among others.

Our Digital Library currently contains a variety of collections, from sources as diverse as images of California natural resources and dams from the State Dept of Water Resources, images from the Fine Arts Museum of San Francisco, California environmental impact reports and plans, vertebrate biology specimen records, and fossil records.

*This project is not officially supported through CITRIS funds, but the faculty and topical affiliations are sufficiently strong that it is listed here for referral and convenience.