ISR11 A Modified Two-Level Search Algorithm Using Request Clustering chapter V. R. Lesser Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. VII-l VII. A Modified [OCRerr]o-Level Search Algorithm Using Request Clustering V. R. Lesser 1. Introduction In the past few years, prototype time sharing computer systems have been developed which have made it possible to obtain access to computers by remote console. In the conte[OCRerr][OCRerr] of an information retrieval system,[OCRerr]this development is likely to affect the systems operations: from a batch type processing of queries to single query processing introduced into the system via remote console. Of still greater importance is the fact that this change makes possible the use of an information retrieval system by a large and diverse user population. Because of these new developments in computer organization, a considerable degree of emphasis has been placed on procedures for using a system of man-machine interaction to improve the retrieval of relevant documents in answer to search requests from a popula- tion of users L[OCRerr],5,6]. Such a change of procedure necessitates a redesign of the techniques of document retrieval to make them adaptable to a single query processing environment. In a batch processing organization, it is not unreasonable to wait until a large set of queries accumulates, and thereafter to search the whole document collection in one pass to identify documents which are highly correlated vTith the batch of queries. In a real time system, on the other hand, queries cannot be batched; as a result a search of the whole document collection for each query becomes very uneconomical, and the