An adaptive information retrieval system using partial file inversion

Abstract

The retrieval of information by means of freely-chosen descriptors may be handled efficiently by creating an inverted file organization in an adaptive manner, entries in this file being generated by sequential searching of the master file, when required. A system design utilizing this concept and incorporating features to generate file keys of wide generality, to minimise sequential searching, and to regulate the size of the inverted file, is described. Operational parameters which can be varied to tune the system for efficient operation are discussed. Some potential applications of this type of retrieval system are presented.