INTERNET RESOURCE DISCOVERY AT THE UNIVERSITY-OF-COLORADO

The Internet is a collection of thousands of networks connecting millions of users at academic, industrial, and government institutions worldwide. This extensive connectivity offers tremendous potential for collaboration and sharing of resources, such as documents, data, network services, and personal expertise. Yet locating resources of interest is difficult because of the scale, administrative decentralization, and constantly changing nature of the environment. In this article, the author discusses aspects of the resource discovery problem: how users specify searches, the difference between discovering classes of resources and locating appropriate instances, system-management problems that can be cast as global state discovery searches, issues involved with characterizing resources and with the efficient distribution of characterizing information, and social issues, especially privacy. The author presents results of efforts to address these problems as part of the Networked Resource Discovery Project at the University of Colorado. These efforts use a variety of experimental approaches, including prototype systems, network measurement studies, and simulation studies. The author provides brief overviews of a half dozen of these efforts: the Netfind Internet user directory service measurements of popular file duplicate transmission rates in the Internet, algorithms for discovering shared interests among people based on graph analysis, simulation results for a probabilistic approach to distributing source information, and a prototype system for assisting users in locating Internet file archives.