Automation of thesauri: Its effect on their structure

Abstract

The capabilities of computer programs affect considerably the characteristics of mechanized thesauri applied in automatic information retrieval systems. Due to automatic reference from nondeseriptors to deseriptors, relations based on real synonyms and quasisynonyms are designated differently. In consequence of the automatic handling of relations between deseriptors, more and semantically more precise relations are used than usual. As a result of using syntactical relations certain compound words are not included in the thesauri, because the unanimity can be assured through post-coordination. Adjectives and other parts of speech are used as deseriptors more often than usual. Automatic indexing results in identical problems. During acquisition, more attention should be paid to the language component of the available programs, because they differ considerably in this respect.