ISR11 The SMART System -- Retrieval Results and Future Plans chapter 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. `-3 the given request. The search is then limited to only those documents included in the previously identified subsets. The query clustering process, on the other hand, depends on the accumulation of groups of requests previously processed through the system. In that case, the search strategy for a given query can be made to depend on the strategies previously found useful for siniilar types of queries. In either case, only a small portion of the collection is actually involved in the search process, [OCRerr]nd the actual loss in search effectiveness, compared ;[OCRerr][OCRerr]th a full search is found to be small. In the next part, a few of the principal evaluation results obtained with the SMART system are summarized, and some of the future research plans are discussed in part three of this section. 2. E[OCRerr]erimental Results The initial experiments conducted [OCRerr][OCRerr]th the SMART system were specifically designeA to answer certain fundamental questions concerning the design of information systems can auto[OCRerr]r'[OCRerr]tic text processing methods be used effectively to replace a manual content analysis; if so, what parts of the documents are - most appropriate for incorporation into the analysis; is it necessary to provide vocabulary normalization methods to elin[OCRerr]nate linguistic ambiguities; should such noz[OCRerr]nalization be handled by means of specially constructed dictionaries, or is it possible to replace thesauruses by statistical word association methods; what dictionaries can be used most effectively for vocabulary normalization; is it important to provide hierarchical subject arrangements, as is done in library classification systems; alternatively,