Massively parallel information retrieval for wide area information servers

Most text is stored in electronic form at one point or another in its file. Computers have made relatively little impact on how it is stored and retrieved. The wide area information service project (WAIS) seeks to alleviate this problem by making text databases widely accessible via a common user interface and wide area networking. The ultimate success of the effort depends, in part, on the availability of high-performance text search engines. Such engines can be implemented on massively parallel machines, using either signature file or inverted file database structures. Inverted files are particularly appealing, and permit databases as large as 8192 Gbytes to be searched in under 15 s, using currently available hardware