The Mead Information Retrieval System

A case study is presented of the LEXIS/NEXIS information-retrieval system, which during peak loads processes nearly eight thousand full-text searches an hour. Most often these searches are directed at databases on the order of one hundred billion bytes in size. The I/O and CPU resource requirements to service the LEXIS/NEXIS workload is approaching the limits of existing computer technology. It is argued that for the information-retrieval industry to remain cost-effective as database size increases, fundamental advances in computer architectures and full-text search algorithms are required.<>