IRS13 Scientific Report No. IRS-13 Information Storage and Retrieval Search Matching Functions chapter E. M. Keen Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. 111-5 documents in the file, and also to the prevention of an exhaustive examination of the file. A manual search may easily be modified as it proceeds to prevent retrieval of more documents than the user is willing to examine, the modi- fication being carried out by controlling the choice of search terms, as well as the logical combinations demanded. B) Mechanized Systems The word T1mechanized" is here used to describe systems in which certain parts only of the storage, processing, and retrieval stages are mechanized. The present discussion thus concerns systems in which the physical search or matching operation is mechanized, but the search formu- lation is manually constructed. Since mechanization of document retrieval systems has been based in the past almost invariably on manual systems of the post-co-ordinate type, search foruulations for mechanized systems require both a selection of search terms and also a request statement in the form of logical conibinations of these search terms. The formulation of one parti- cular search request, together with the requestor's original statement, is given in Figure 2, taken from the Medlars system, in which a file ot nearly half a million items is searched by means of a computer. Systems which mechanize the matching process in this way have solved some of the problems associated with manual systems, but have introduced other problems which are unlikely to be solved by this approach. Three areas of difficulty may be outlined: 1. Both manual and mechanized systems of the type described require manual search formulation, consisting of a choice of terms on which to search, and a statement of the logic that is to isolate a set of retrieved documents. This process takes skill and is time-consuming; good results are not, however, obtained for every search. 2. Although mechanized systems are successful in saving some of the time and effort that is required to maintain the document file,