Effective information‐searching strategies without “perfect” indexing

Abstract

AbstractComputer searching of an information file offers significant advantages over manual searching when it can utilize strategies that are impractical for a human searcher to use. This paper examines several existing strategies, and then suggests that increased emphasis be placed on one that is relatively tolerant of “imperfect” indexing. Since “perfect” indexing does not exist, such a strategy would have considerable value. The general procedure consists of counting the number of search‐matched index terms and then calculating a relevance factor for each item found, thus enabling the final search‐output listing to be prepared in order of probable relevance. This technique is essentially the same as the strategy used successfully in such diverse tasks as identifying criminal suspects by their modus operandi, selecting suitable gifts for department store shoppers, and locating misspelled names in an airlines reservation file.