Adaptable and Efficient Information Retrieval Clients and Servers

Ronald F. Boisvert, ACMD
Jack Dongarra, University of Tennessee at Knoxville
Eric Grosse, AT&T Bell Laboratories

We have begun participation in a project to address limitations and inadequacies of current information retrieval tools by developing new mechanisms for improving the flexibility and efficiency of network client and server applications. This work is motivated by the special needs encountered in the development and maintenance of the Netlib and GAMS software repositories. The following tasks are underway.

1. Location Independent File Name (LIFN) extension to Universal Resource Names (URNs) to allow management of multiple versions of objects.
2. Techniques for the automatic replication and registration of network resources.
3. Safe execution environments for downloaded client-side extension modules.
4. Improved classification technology for software resource identification.

NIST participation includes maintenance of experimental name-to-location lookup and file replication services, as well as demonstration of sophisticated client-server extension modules. We have also begun work on a successor to the GAMS Problem Classification system. The new system is to be a hybrid of hierarchical and faceted schemes which will lead to both precise classification and new modes of user interaction. Broad input to the development of the taxonomy is being sought.