IRS13 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. hI-i III. S[OCRerr]arc[OCRerr] [OCRerr]atchin[OCRerr] [OCRerr][OCRerr]nctions Ee M. Keen 1. Introduction It is the function of a document retrieval system to draw to a [OCRerr] attention some or all of the documents in the store that have some probability of being relevant to the requestor1s need. The prGcessing of a search request is in part a matching operation, and procedures used in manual and mechanized systems are briefly described before automated s[OCRerr]stems are discussed. SMART test results are given, separating the results of the matching functions themselves from the weighting scheme that may be employed. The analyses show that one particular matching function used [OCRerr]ith the weighting scheme is superior to the others that have been tested. A suggested experiment is proposed to determine whether development of still better functions is possible. 2. Matching [OCRerr]ocedures used in Manual, Mechanized and Automated Systems A) Manual Systems The task of matching a search request against a stored file of document representations is frequently an entirely manual process. After the search request has been received and "translated'1 into the vocabulary of the system, either words or code nun[OCRerr]bers, a searcher then seeks documents that match the coded request by referring to the index file, that is, the cards or the pages constituting the physical index. In searches made with dic- tionary or classified catalogs, the searcher normally proceeds by starting