ISR11 Design Criteria for Automatic Information Systems chapter M. E. Lesk G. Salton Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. V-32 may be expected to decrease in effectiveness, because of the variabilities among indexers, and the difficulties of ensuring a uniform application of a given set of indexing rules to all documents. The computer process will, however, not decay as the collections grow larger, and one may anticipate for large collections of operational size an even greater difference in performance, and a clearer advantage for the automatic process. G) Iterative Searching Most presently operating information systems perform a single search operation for each search request, and the user of the system must submit a completely new request if he is dissatisfied with the initial response. This situation is not ideal, since it assumes that a single information analysis and search method[OCRerr]will prove equally useful to all customers, and furthermore that all users have the same type of need and will thus be satisfied with the same type of answer. In actual practice, users have many different needs, some wanting very exhaustive answers, others being content with a single reference. This situation is well recognized, and it is widely felt that the new computer time-sharing organizations, which permit a multiplicity of users to obtain access, more or less simultaneously, to a central equipment complex can be used advantageously to provide individualized service to each customer according to his need. Accordingly, several iterative search methods have been simulated with the SM[OCRerr]T programs. [6,7] In each case, a user first obtains some output in response to an initial query, and depending on what he learns from this output, he returns enough information to the system to permit a reprocessing of the original query